Google says new projects that use the Generate Content API (generateContent, streamGenerateContent and batchGenerateContent) on the paid tier now store requests by default (store=true), matching the Interactions API. Existing projects keep their current setting. To opt out, set store=false on individual requests or turn off Generate Content API logging in the Settings panel of the Logs and Datasets dashboard.
Who can use it
Developers creating new projects on the paid tier of the Gemini API. Existing projects keep their current setting.
When
Listed in the Gemini API release notes on October 26, 2026. No time of day was given.
OpenAI released GPT-6.1 Terra, a fast, efficient model that combines detailed visual reasoning and precise computer use with clearer, more natural writing. In ChatGPT Work and Codex, OpenAI says the same usage allowance covers more tasks than with GPT-5.6 Terra, and your plan's overall usage limits are unchanged. Developers can use it through the OpenAI API as gpt-6.1-terra. Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens, which OpenAI says is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol. Some Plus and Pro users may see GPT-6.1 Terra at Medium instead of GPT-6.1 Sol at Light in the default Power presets in Work.
Who can use it
ChatGPT Work and Codex on Plus, Pro, Business, Enterprise and Edu plans, as it reaches your account; plan, region and workspace settings apply. The launch article names Plus, Pro, Enterprise and Edu. Not in Chat. No UAE inference residency. Also in the API.
When
Announced October 26, 2026. The launch article says 'Starting today'; the release notes call it a rollout, so it may not have reached your account yet. No time of day was given.
Anthropic released the Claude for Nonprofits plugin, a set of skills and connectors built with about 30 nonprofits. It covers fundraising, from finding funders and grants to writing proposals and reports, plus marketing, hiring and people, and board and leadership work. Setup skills map where an organization keeps its records and let staff check what its AI policy allows. Anthropic says this plugin is the first of several releases built with nonprofit organizations, and that it plans to extend Claude Corps through a partnership with Taproot.
Who can use it
Every verified nonprofit organization. The plugin is installed automatically for nonprofits on Team and Enterprise plans, at no additional cost.
OpenAI's Codex lead, Tibo Sottiaux, posted: “Day 17/ Bet you thought we forgot, but no, it’s reset day today.” The post does not say which plans get the reset or what it covers. A follow-up post says: “Lands tomorrow before 10am PST.”
Correction: Updated 23 Oct 2026: the reset is no longer listed as expected. The timing now quotes the source, “before 10am PST”, instead of a converted UTC time.
Who can use it
Not specified by the source
When
Announced October 22, 2026 at 06:55 UTC. A follow-up post says “Lands tomorrow before 10am PST.” The posts give no exact landing time.
Mistral says Z.ai GLM 5.3 Flash (zai-glm-5-3-flash) is now available. It is a Z.ai model with a 1M-token context window. Pricing: $0.15/M input tokens, $0.03/M cached input, $0.50/M output tokens.
Who can use it
Available on Mistral.
When
Available October 22, 2026. The source gives a date, not an exact time.
Anthropic launched Claude Models on Google Cloud (Anthropic managed), the Claude API operated by Anthropic and reached through your Google Cloud project. Google Cloud IAM controls access, and Google Cloud Marketplace handles billing. Fast mode is available for Claude Opus 5.5 on Claude Models on Google Cloud (Anthropic managed).
Who can use it
Developers reaching the Claude API through their Google Cloud project, with access through Google Cloud IAM and billing through Google Cloud Marketplace.
When
Generally available October 22, 2026. The source gives a date, not an exact time.
OpenAI says the Decisions API (v1/decisions) is now generally available with gpt-6-luna. Each request now supports up to 400 questions, increased from 200 in public beta.
Who can use it
OpenAI API developers, with gpt-6-luna.
When
Generally available October 22, 2026. The source gives a date, not an exact time.
Anthropic opened a research preview of Claude Duet, a new app where Claude works as a proactive agent. Anthropic says it comes to you with what needs your attention, completes tasks you ask of it, keeps working even when you step away, and checks with you before it acts on your behalf. It is available as a standalone app on iOS and Android, and on claude.ai. Access is by waitlist, starting with US Pro and Max subscribers. For the research preview, Claude Duet is included in Pro and Max plans and uses the same usage limits as the rest of Claude.
Who can use it
By waitlist, starting with US Pro and Max subscribers. Included in Pro and Max plans during the research preview.
When
Research preview opened October 21, 2026, by waitlist. The source gives a date, not an exact time.
Anthropic released API version 2026-10-21. In this version the sampling parameters temperature, top_p and top_k are removed and return a 400 error. New versions of the client SDKs and the ant CLI send API version 2026-10-21 by default: Python SDK 2.0.0, TypeScript SDK 1.0.0, C# SDK 13.0.0, Go SDK 2.0.0, Java SDK 3.0.0, PHP SDK 1.0.0, Ruby SDK 2.0.0 and ant CLI 2.0.0. Anthropic says these SDK and CLI versions also have breaking changes that don't depend on the API version, and are licensed under Apache 2.0 instead of the MIT License.
Who can use it
Developers calling the Claude API with API version 2026-10-21, including through the new SDKs and CLI that send it by default. Earlier API versions return the error only for non-default values on Claude 4.7 and later models and Claude Mythos Preview.
When
Released October 21, 2026. The source gives a date, not an exact time.
OpenAI is expanding ChatGPT Ads to 26 additional countries, extending availability to 89 countries worldwide. The new markets include 14 more Latin American countries such as Argentina, Chile, Colombia and Peru, plus Nigeria, South Africa, Serbia, Ukraine, Cambodia, Bangladesh and Pakistan. Ads appear only on the Free and Go plans; Plus, Pro and Enterprise stay ad-free. OpenAI says over 100,000 advertisers have run campaigns since ads launched just over eight months ago.
Who can use it
Ads appear only on the Free and Go plans in the listed countries. Plus, Pro and Enterprise subscriptions remain ad-free. Advertisers use technology partners or self-service Ads Manager.
When
Announced October 21, 2026. The expansion is described as launching today.
OpenAI's Codex lead, Tibo Sottiaux, posted: “Day 15/ Nevermind, we're still cooking... and reset. Enjoy.” The post quotes his earlier one, “Day 15/ We're cooked”. The post does not say which plans get the reset, what it covers or when it lands.
Who can use it
Not specified by the source
When
Announced October 20, 2026 at 06:33 UTC. The post does not say when the reset lands.
Google said Gemini is now available globally on Android Auto and cars with Google built-in, so drivers can speak to it naturally. Personal Intelligence, which with permission connects Workspace (including Gmail and Google Calendar), Google Photos, Google Maps, YouTube and Search, starts rolling out on Android Auto. It reaches cars with Google built-in in the coming months. It is off until the user turns it on in Gemini settings. Plan and pricing details were not stated.
Who can use it
Personal Intelligence is rolling out to eligible personal Google accounts, not Workspace business, enterprise or education users, aged 18 and older globally, excluding the European Economic Area, Switzerland, the United Kingdom and Nigeria.
When
Announced October 20, 2026. Personal Intelligence rolls out from that day on Android Auto and in the coming months on cars with Google built-in.
Mistral released the Voxtral 3 family: Voxtral Mini Transcribe 3, Voxtral Mini Transcribe Realtime 3, Voxtral Small Transcribe 3 and Voxtral TTS 3 (voxtral-mini-tts-3-0), plus a voice agent SDK built on LiveKit that runs speech in, reasoning and speech out as one streaming pipeline. All four models are in Public Preview. Listed prices are $0.0035 per minute for Mini Transcribe 3 (batch), $0.008 per minute for Mini Transcribe Realtime 3, $0.0040 per minute for Small Transcribe 3 and $13 per 1M characters for TTS 3. Open-weights availability was not stated in the announcement.
Correction: Corrected 23 Oct 2026: added that all four models are in Public Preview, and named the text-to-speech model as Mistral's docs do: Voxtral TTS 3 (voxtral-mini-tts-3-0).
Who can use it
Available through the Mistral API on European, US and global endpoints, with playgrounds in Studio. On-premises or private-cloud deployment is by contacting Mistral.
Anthropic released the Claude Tag API in beta for Claude Enterprise organizations, using the experimental-claude-tag-2026-10-21 beta header. Its /v1/organizations/claude_tag/ endpoints read and change Claude Tag settings on Slack channels and Slack workspaces, show which settings apply in a channel and where each comes from, and attach Access bundles. It needs an Admin API key with the read:claude_tag or write:claude_tag scope.
Who can use it
Claude Enterprise organizations. It is not available to Claude Platform (Claude Console) organizations.
When
Released in beta on October 20, 2026, per Anthropic's release notes, which give a date and no time.
Google said Gemini is launching a central hub to help small business owners manage tasks, finances and marketing in one place. Users can connect existing apps and upload brand guidelines to generate consistent visuals automatically. Plan and pricing details were not stated.
Who can use it
Small business owners using the Gemini app. Plan requirements were not published.
When
Announced October 20, 2026. Google said the tools roll out globally in the coming days.
OpenAI added a 12-hour prompt caching option to the Responses API. It keeps shared instructions, tools and context cached longer to benefit from lower cached-input pricing, even when requests are hours apart. 12-hour cache writes cost 2.5x the input token price. The changelog did not publish the cached-input rate.
Who can use it
API developers using the v1/responses endpoint. Model coverage was not stated in the changelog entry.
OpenAI introduced locked chats in ChatGPT. A locked chat stays in your account but is hidden from regular chat history, pinned chats and search. You open locked chats with your 4-digit account PIN or supported device authentication, and your retention and model-improvement settings stay unchanged. To lock a chat, open its more options menu and select Lock chat.
Who can use it
Rolling out gradually to Free, Go, Plus and Pro users on the web, in the iOS and Android apps, and in the new ChatGPT desktop app for macOS and Windows.
When
Announced October 20, 2026 and rolling out gradually. OpenAI gave no completion date.
OpenAI's Codex lead, Tibo Sottiaux, posted “Day 13/ Le limit reset” with an image of a card reading “Le Limit Reset”. The post does not say which plans get the reset or what it covers. A follow-up post says: “Landing sometime today by EOD PST”.
Who can use it
Not specified by the source
When
Announced October 17, 2026 at 16:20 UTC. A follow-up post says “Landing sometime today by EOD PST”.
DeepSeek says V4.1-Flash-1017 has stronger benchmark performance, faster task completion and shorter outputs than the previous version. It is live on DeepSeek's website, app and API, and the weights are published on Hugging Face and ModelScope. The post does not give benchmark numbers or API pricing for this model.
Who can use it
Live on DeepSeek's website, app and API. Regional or account restrictions are not specified by the source. DeepSeek recommends upgrading to DeepSeek Harness v0.2.1-rc.1 or later for the best experience.
When
Announced October 17, 2026 and live at announcement.
OpenAI updated GPT-6 Pro in the ChatGPT chat tab. It now offers faster responses, improved use of plugins and tools to carry out tasks, and multiple agents that can work on different parts of a task in parallel. Users can follow the agents' progress in the conversation and open completed tasks to see what each agent did. OpenAI did not publish speed figures.
Who can use it
Rolling out to eligible Pro users in ChatGPT. Availability in managed workspaces depends on workspace settings and rollout eligibility.
When
Announced October 16, 2026 and rolling out now. OpenAI gave no completion date.
OpenAI added ChatGPT Slides, which creates presentations and lets a team keep working on them in Space. You can start from a conversation, files or an outline, or share an example deck to guide style and layout. Text, images, shapes and charts can be edited directly or by asking ChatGPT, and teammates can edit the same deck in real time and leave shared comments. Decks can be presented from ChatGPT; on the web they export to PowerPoint, PDF or Google Slides, and the desktop app can open a deck in PowerPoint or Keynote when installed. OpenAI did not publish usage limits for Slides.
Who can use it
Eligible Pro, Business and Enterprise users with Space access, on the web and in the desktop app.
When
Rolling out from October 16, 2026. OpenAI did not publish a completion date. To start, ask ChatGPT in Work to create a presentation in Space.
OpenAI released prompt-to-voice for all API customers. A custom voice can be created from a text description through the API or the voice Playground and reused in GPT-Live sessions. Creating a prompt-generated voice is free; speech generation uses regular GPT-Live session billing. OpenAI also added nine natural voices for GPT-Live: Nuri, Juni and Flitz (German); Brise and Sillage (French); Nira (Hindi); Harema and Shida (Japanese); and Noeul (Korean).
Who can use it
All API customers (prompt-to-voice, gpt-live-1, v1/audio/voices and v1/live/sessions).
When
Released October 16, 2026. OpenAI did not publish a rollout schedule.
OpenAI's Codex lead, Tibo Sottiaux, posted: “The reports of my reset button being lost are greatly exaggerated” and “Day 10/ Reset”. The post does not say which plans get the reset or what it covers. A follow-up post says: “Landing within the next 2 hours”.
Who can use it
Not specified by the source
When
Announced October 15, 2026 at 04:55 UTC. A follow-up post 16 seconds later says “Landing within the next 2 hours”.
OpenAI's ChatGPT release notes for October 15, 2026 say it is rolling out GPT-Live-1.1 and GPT-Live-1.1 Quip, a mini version of the model, in ChatGPT Voice, with more natural accents across languages. Live also uses a wider range of short phrases to let you know when it is searching. New Live voices include options for French, German, Hindi, Japanese, Korean and Mexican Spanish, plus Bubbie in English. Usage limits and plan availability were not published in the note.
Who can use it
ChatGPT Voice users. OpenAI says the voices you see may vary by region and account; which plans get the models was not published.
When
Rolling out from October 15, 2026. No completion date was published.
Mistral announced Timestral on October 15, 2026 as its first family of time-series foundation models, available in four sizes from 5M to 209M parameters. Mistral says the largest model, at 209M parameters, ranks first on the GIFT-Eval, TIME and fev-bench public benchmarks, and that the models produce probabilistic, multivariate forecasts and support past and known-future covariates. Mistral says Timestral is available through Mistral Forge. API pricing and open-weights availability were not published in the announcement.
Who can use it
Organisations using Mistral Forge. Pricing and any other access route were not published.
Anthropic says Code Review, which reviews pull requests automatically in Claude Code, is out of research preview and generally available to Claude Team and Enterprise organizations. Anthropic says most reviews now finish about 3x faster than in the preview, in 6-10 minutes instead of about 20, and admins can choose a Quick, Standard (default) or Deep review level for each repository. Eligible Enterprise organizations can claim at least $10,000 in Code Review credits by January 15, 2027; credits expire 30 days after they are claimed and terms apply.
Who can use it
Claude Team and Enterprise organizations. Admins enable Code Review in Claude Code settings, install the GitHub App and choose repositories; reviews then run automatically on new pull requests. The credit offer is for eligible Enterprise organizations only.
When
Generally available from October 15, 2026. Enterprise credits must be claimed by January 15, 2027 and expire 30 days after claiming.
OpenAI introduced the Rosalind Academic Plan for academic life sciences labs. Eligible labs can receive up to 30 ChatGPT Business Standard seats at no cost for up to 12 months, with access to all commercially available frontier OpenAI models subject to plan usage limits. Access to GPT-Rosalind, OpenAI's life sciences model, depends on trusted-access approval. Applying does not give a lab immediate access.
Who can use it
Applications are open to labs at U.S. universities, submitted by a principal investigator or lab lead on behalf of the lab. Labs covered by an OpenAI Enterprise agreement should use their university's access process.
When
Announced October 15, 2026 and applications are open now. Seats run for up to 12 months after approval. OpenAI has not published a closing date for applications.
Anthropic says fast mode is out of research preview for Claude Opus 5.5 and you no longer need to request access. Requests no longer require the fast-mode-2026-02-01 beta header, and requests that still send it keep working. Claude Opus 5.5 now has a default fast mode rate limit in the Start, Build, and Scale usage tiers; the limit numbers were not published in the note. Fast mode is also available for Opus 5.5 on Claude in Microsoft Foundry for deployments hosted on Azure.
Who can use it
Claude API users on Claude Opus 5.5 in the Start, Build, and Scale usage tiers, plus Claude in Microsoft Foundry deployments hosted on Azure. Fast mode for Claude Opus 5 and Claude Opus 4.8 remains in research preview, and requests to those models must keep sending the beta header.
When
Announced October 15, 2026 in the Claude Platform release notes. Applies now.
xAI announced Grok Voice 3.0 on October 14, 2026 and says it is available today in the Voice Agent API at $0.08 per minute ($4.80 per hour), the same price as Grok Voice 2.0. The grok-voice-latest alias has already been switched to grok-voice-3.0, so agents using the alias are upgraded automatically with no changes needed. xAI's release notes say grok-voice-2.0 is deprecated and give no shutdown date. xAI reports an Artificial Analysis Speech-to-Speech Index score of 86.9 and says Grok Voice 3.0 also powers Grok Bot's voice mode.
Who can use it
Developers using the xAI Voice Agent API (Speech to Speech). Agents on grok-voice-latest are moved automatically; agents pinned to grok-voice-2.0 need to set the model to grok-voice-3.0 or grok-voice-latest.
When
Available from October 14, 2026. No shutdown date has been published for grok-voice-2.0.
Meta introduced Muse Wonder and Muse Wonder Full, which it calls its most capable models to date, built for personal and professional agentic tasks in areas including law, finance and health, plus coding. Muse Wonder is available today to developers and builders through the Meta Model API and Muse Code. Muse Wonder Full is available today to Muse Code subscribers. Meta says Muse Wonder will soon power more of Muse, rolling out first to subscribers. Pricing and usage limits were not published in the post. A full safety report is described as coming soon.
Who can use it
Muse Wonder: developers and builders via the Meta Model API and Muse Code. Muse Wonder Full: Muse Code subscribers. Muse app: coming soon, subscribers first. Pricing, regions and limits: not specified by source.
When
Announced October 12, 2026 (newsroom page also dated October 13). Muse app rollout is 'soon' with no date.
xAI introduced The X Subscription, which combines Grok, Grok Bot, Cursor and X Premium in one plan with a shared pool of Grok model usage. Plans are Free ($0), Plus ($8/month), Premium ($30/month), Super ($100/month) and Ultra ($200/month); Premium is described as replacing SuperGrok, Cursor Pro and X Premium+, previously $90/month combined. xAI says each plan includes at least twice as much coding usage with its models as the comparable Cursor plan. It is available to new customers starting today and is being made available to everyone globally over the next few days. Existing SuperGrok, Cursor and X Premium plans continue unchanged, with an option to switch.
Who can use it
New customers from October 12, 2026; global availability over the next few days. Existing SuperGrok, Cursor and X Premium subscribers can keep their plan or switch in billing settings when the option appears for their account. Teams and Enterprise plans are offered.
When
Announced October 12, 2026. Old plans are closed to new sign-ups; xAI says it will eventually move remaining customers but gives no timeline.
Qwen Cloud lists qwen-image-2.1-turbo and describes it as the most balanced and cost-effective open-source image generation model currently available in the Qwen-Image series. The changelog lists four strengths: compact model, strong performance and value; native transparency with unified generation and editing; versatile editing with multiple references, local control and fidelity; and realistic textures, typography and portraits. Pricing and limits were not published in the entry.
Who can use it
Developers using Qwen Cloud (model name qwen-image-2.1-turbo). Plans, regions and pricing: not specified by source.
When
Listed in the Qwen Cloud model changelog on October 9, 2026.
Anthropic launched the Anthropic Cyber Mission, a long-term effort to support defenders with tools, research and resources, starting with critical infrastructure and open-source software. The Critical Infrastructure Defense Program brings frontier models, on-site engineers and threat research to the defenders that protect operational technology; its founding partners include Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Insane Cyber, Nozomi Networks, Palo Alto Networks, PwC and Rockwell Automation. OSS Scanner offers open-source projects regular security scans from Anthropic's strongest models, for free. Usage limits and pricing were not published.
Who can use it
Critical Infrastructure Defense Program: a small cohort of providers to start; companies that build security products or services for critical infrastructure can register interest. OSS Scanner: open-source projects.
When
Announced October 8, 2026. Anthropic says the Critical Infrastructure Defense Program work is underway with several partners; no wider date was given.
OpenAI added Ultrafast mode for GPT-6.1 Sol in the Responses API. Developers use gpt-6.1-sol with service_tier set to "ultrafast" to reduce the time between generated output tokens. It has global processing and US and EU data residency. OpenAI's changelog points to separate Ultrafast pricing; the price is not stated in the changelog entry.
Who can use it
All API users, subject to rate limits.
When
Announced October 8, 2026 in the OpenAI API changelog. Available now.
Anthropic launched Claude Dashboards, which connects to a data platform such as BigQuery, Databricks or Snowflake, or a CRM tool such as Salesforce, and builds a dashboard that stays current as the data changes. It also launched Claude Motion, which builds editable animated explainers as code, plays them like a short video and exports them as MP4 files. Claude Docs, Slides and Design are out of beta and available on every plan, including Free. Anthropic did not publish usage limits or pricing in the announcement.
Who can use it
Claude Dashboards: beta on paid plans. Claude Motion: beta on Team and Enterprise. Claude Docs, Slides and Design: every plan, including Free.
When
Announced October 8, 2026, starting today. Both new features are in beta. Anthropic lists Looker, monday.com and Tableau as coming soon for sending dashboards to analytics tools.
ChatGPT's Chat tab is moving to GPT-6 with Intelligent UI, which lets ChatGPT answer with interactive elements alongside text, such as side-by-side comparisons, diagrams, calculators and bill splitters. Plus, Pro, Business and Enterprise use GPT-6 Sol; Free and Go use GPT-6 Luna. GPT-6 can also start answering while it is still thinking or using tools. Intelligent UI works from Instant through Extra High reasoning; the Pro reasoning option stays on GPT-6 Astra without Intelligent UI. The models behind Work and Codex are not changing.
Who can use it
Chat tab only. Plus, Pro, Business and Enterprise get GPT-6 Sol; Free and Go get GPT-6 Luna. Enterprise availability depends on workspace admin settings.
When
Announced October 7, 2026. Rolling out globally to paid tiers from October 7 and to Free and Go from October 8.
Anthropic halved the price of prompt cache reads on Claude Sonnet 5.5, from $0.20 to $0.10 per million tokens (0.05x the base input price instead of 0.1x). Cache writes and all other prices are unchanged. Anthropic says this cuts the cost of Sonnet 5.5 on most agentic tasks by around 20%.
Who can use it
Developers using prompt caching with Claude Sonnet 5.5, per the Claude Platform pricing.
Claude Max and Team subscribers now get monthly credits for the Claude API: $100 on Max 5x, $200 on Max 20x, and on Team $20 per Standard seat and $100 per Premium seat, pooled up to $500. To claim them, you link a Claude Console organization to your plan; no payment method is needed. Credits refresh each billing cycle (monthly for annual plans) with no rollover, and they work only on the Claude API in the Claude Console, not on other platforms.
Who can use it
Max 5x, Max 20x and Team, including discounted Team plans. New subscribers can claim after 7 days on an eligible plan. Free, Pro and Enterprise aren't eligible.
Anthropic released Claude Haiku 5.5, which it calls its cheapest, fastest and most capable small model. It is priced by prompt length: $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, and $0.50 / $2.50 above that. Anthropic says it costs around 75% less to run than Haiku 4.5 on average. It has a 1M-token context window, 128k max output tokens and adaptive thinking with the effort parameter, making it the first Haiku-class model with an adjustable effort setting. Code written for Haiku 4.5 can break: manual extended thinking (budget_tokens) returns a 400 error.
Correction: Updated on 8 Oct: Claude Haiku 5.5 is now available. Pricing, limits and platforms come from Anthropic's 7 October launch.
Who can use it
Developers on the Claude API (claude-haiku-5-5), Claude in Amazon Bedrock, Claude Platform on AWS, Claude on Google Cloud and Claude in Microsoft Foundry.
When
Available from October 7, 2026. First announced on September 22, 2026 as coming 'in the coming weeks'.
Anthropic released Claude for Google Workspace in public beta on all paid Claude plans. The add-on opens Claude in a sidebar next to your Google Docs, Sheets or Slides file, where it can read the open file, see what you have selected and make changes directly in it. In Docs it edits in place or proposes edits as suggestion cards; in Sheets it writes formulas, builds pivot tables and native charts and can pull a range into Python; in Slides it builds new slides from the deck's layouts and themes. A default 'Ask before edits' mode shows each change as an approval card, and an 'Accept all edits' mode applies changes without stopping. Signed-in users get the same models, connectors and skills they use in Claude. Anthropic is also releasing Google Docs, Sheets and Slides connectors (beta) so Claude can create and edit Google files from a Claude chat; access matches your existing Google sharing permissions.
Who can use it
All paid Claude plans. Install from the Google Workspace Marketplace; admins can deploy it to their domain or selected groups from the Google Admin console. Enterprise controls such as the Compliance API, CMEK and OpenTelemetry audit export apply to the add-on.
When
Announced October 6, 2026 and available in public beta the same day. The Docs, Sheets and Slides connectors are also beta. No general-availability date is given.
Google released Gemini Nano Banana 2.1 (gemini-nano-banana-2.1) in the Gemini API as generally available. Google calls it its latest high-efficiency image generation and conversational editing model. It is an update to Nano Banana 2 (gemini-3.1-flash-image) that maintains Flash-level speed and cost efficiency while, Google says, improving visual quality, prompt adherence, multi-turn character consistency and text rendering, and adding wide and panoramic aspect ratios (1:4, 4:1, 1:8, 8:1) across 1K, 2K and 4K resolutions. In the same note Google deprecated gemini-3.1-flash-image, with no shutdown date announced, and tells developers to migrate to gemini-nano-banana-2.1.
Correction: Corrected 22 Oct 2026: Google has not announced a shutdown date for gemini-3.1-flash-image. An earlier version of this page said 29 October 2026, which was wrong.
Who can use it
Gemini API developers. The release note covers the API only and does not say whether the model is in the Gemini app or give pricing.
When
Released October 6, 2026 in the Gemini API. Google deprecated the predecessor gemini-3.1-flash-image the same day, with no shutdown date announced.
OpenAI added audio uploads to ChatGPT. You can upload audio files to create transcripts, summarize recordings and ask questions about their contents, for example turning a meeting, interview or lecture into structured notes or a follow-up email draft. To start, attach a supported audio file to a conversation and describe what you need. OpenAI notes that transcripts may contain errors and performance may vary across languages.
Who can use it
Paid ChatGPT subscriptions and workspaces. OpenAI says availability may vary by workspace settings, region, client version and model.
When
Listed in the ChatGPT release notes on October 6, 2026. No staged rollout dates are given.
Anthropic launched an expanded Cyber Verification Program (CVP), which makes advanced cyber capabilities and reduced blocking classifiers available to qualifying security professionals. It now has three access tiers. Defense Access covers defensive work such as security operations, incident response, malware reverse-engineering and vulnerability validation; Anthropic expects many defensive organisations to qualify and aims to respond within a few days. Red Team Access adds authorised penetration testing and red-teaming for organisations only, with reviews expected to take a few weeks. Specialized Access, with the fewest cyber blocks, is reserved for a limited set of verified organisations that test safety-critical systems, reviewed in depth with the US government. Each tier includes Anthropic's most capable models, including Claude Opus 5.5, Claude Sonnet 5.5 and Claude Mythos 5.1. Project Glasswing folds into Specialized Access, and existing members transition without reapproval for current models. Generally available models keep their conservative cyber safeguards.
Who can use it
Qualifying security professionals and organisations who apply. Defense Access includes individual researchers with a track record; Red Team Access is for organisations only; Specialized Access is a limited set of verified organisations.
When
Announced October 6, 2026; applications open the same day. Anthropic aims to answer Defense Access applications within a few days and Red Team Access within a few weeks.
Mistral launched a public preview of Mistral Large 4 (ML4), which it nicknames le Chonk. The preview API is available today on Mistral Studio. Mistral describes ML4 as a 1 trillion-parameter natively multimodal model with 49 billion active parameters, its largest and most capable model to date, trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in its own datacenters in Europe. Mistral says it will release the weights by the end of the month. Until then it is red-teaming the model with cybersecurity leaders, vetted partners and state authorities, who get the same model with reduced moderation and expanded cyber capabilities. Mistral says the model will be available across multiple regions worldwide, including a European deployment it operates end-to-end.
Who can use it
Anyone with Mistral Studio access can try the preview API today. Open weights are promised but not yet released. Reduced-moderation access during the preview is limited to red-teaming partners.
When
Announced October 6, 2026; the preview API opened the same day. Mistral says the weights drop by the end of the month (October 2026) and gives no exact date.
OpenAI released the Decisions API (v1/decisions) in beta with gpt-6-luna. OpenAI says it turns text and images into typed answers 10x faster than the Responses API. The changelog entry gives no pricing or rate-limit details.
Correction: Updated 23 Oct 2026: OpenAI made the Decisions API generally available on October 22, 2026, so the timing no longer says no general-availability date is given.
Who can use it
OpenAI API developers. The beta launches with gpt-6-luna; no other models are named.
When
Released in beta on October 6, 2026. OpenAI made it generally available on October 22, 2026.
API customers worldwide can now opt in to invisible text watermarking for select models; it stays off by default. Over the coming weeks, OpenAI will add an invisible watermark to eligible ChatGPT and Codex text output for users on all plans in the EU only; it is not a global default. OpenAI is taking applications from approved researchers and expert organizations for its text watermark detector, which is not public.
Who can use it
API: customers worldwide, opt-in, select models (not named). ChatGPT and Codex: eligible users on all plans in the EU. Detector: approved researchers and expert organizations, case by case.
When
Announced October 5, 2026. API opt-in available now; EU ChatGPT and Codex watermarking over the coming weeks, with no date given.
OpenAI's Codex lead, Tibo Sottiaux, posted: “Global reset landing tomorrow 10am PST for all paid ChatGPT accounts. Apologies for the slow start with GPT-6.1 Sol, it's now back to running at expected speeds after the massive load spike in the first two days.” He later posted: “Reset all propagated. Enjoy.”
Who can use it
All paid ChatGPT accounts, per the post.
When
Announced October 2, 2026 at 02:14 UTC: “Global reset landing tomorrow 10am PST”. Confirmed “Reset all propagated” at 21:18 UTC the same day.
Qwen Cloud lists qwen-image-2.1-pro, a unified text-to-image and image editing model. Its visual generation component has 7 billion parameters. It natively supports transparency: the model decides from your prompt whether to output a regular image or one with an alpha channel. Image editing accepts up to 10 reference images.
Who can use it
Developers using Qwen Cloud (model name qwen-image-2.1-pro). Plans, regions and pricing: not specified by source.
When
Listed in the Qwen Cloud model changelog on October 2, 2026.
Finances in ChatGPT is rolling out to Free and Go users in the U.S. on web, iOS and Android. OpenAI says you can securely connect your financial accounts to find ways to save, understand your spending and make sense of your investments, with answers grounded in your finances.
Who can use it
Free and Go users in the United States, on web, iOS and Android. The release note does not mention other countries.
When
Listed in the ChatGPT release notes on October 2, 2026 as rolling out. No completion date is given.
Anthropic introduced mods, small TypeScript functions that change how Claude Code behaves. A mod can rewrite a prompt, add interface elements, replace a built-in feature or add new functionality. Mods ship inside plugins, so they are installed and shared like any plugin. Anthropic warns that mods are not sandboxed and run with the same access to your machine as Claude Code, so install only from sources you trust. The built-in /diff feature is now a mod.
Who can use it
Claude Code CLI and desktop app users, via plugins that include mods. Existing plugin controls apply to admins; Team and Enterprise plans load a built-in 'sec-default' mod first. Other plan limits: not specified by source.
When
Available from October 1, 2026, in the Claude Code CLI and desktop app.
OpenAI added two shopping features to ChatGPT. A 'Try on' button appears on product listings for clothes and accessories; select it, then take or upload a selfie to generate a virtual try-on with ChatGPT Images. Your reference photo is saved for future try-ons and can be changed or deleted in Settings, Personalization, Reference photos. You can also save any product to Favorites or create a folder to organize your finds in your ChatGPT Library.
Who can use it
Available in ChatGPT on mobile and web. The release note does not restrict it to particular plans or regions.
When
Listed in the ChatGPT release notes on October 1, 2026.
Google is rolling out skills in Gemini chat globally. A skill saves your instructions so you can reuse them by typing a forward slash and its name. You can stack several skills in one prompt and, from today, add reference files such as text documents, PDFs or images. Skills will replace Gems: Google says it will automatically migrate Gems into skills when Gems go away. Opal, its mini-app experiment, also turns down in November; Google says Gems by Google Labs will turn down too and will not migrate into skills.
Who can use it
Gemini chat: all Google AI subscription tiers, currently 18+ (under-18 'coming soon'). Google Workspace business, enterprise, nonprofit and education customers: 'in the coming weeks'.
When
Rolling out from September 30, 2026. Gems support ends from November 2026 (personal accounts), March 2027 (Workspace business, enterprise, nonprofit) and June 2027 (Workspace education).
Anthropic made Claude for Government generally available for federal and state agencies. The platform delivers Claude's coding and agentic work capabilities through a FedRAMP High authorized environment and has been in public beta since July. Agencies pay for usage in fixed increments with a hard not-to-exceed cap, with no seat fees. The Claude Code command-line interface and Claude for Microsoft 365 are also rolling out in early access.
Who can use it
Federal and state agencies; new agencies request access at claude.com/solutions/government. Claude Code CLI and Claude for Microsoft 365 early access: contact Anthropic's public sector team.
When
Generally available from September 30, 2026. Early access for Claude Code CLI and Claude for Microsoft 365 has no stated date.
Google says Gemini 4 Argon is now globally available to developers and enterprises through the API in AI Studio and the Gemini platform, to Google AI Ultra subscribers in Google Antigravity, and to Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search. The Gemini API changelog lists gemini-4-argon as generally available. The model launches at an introductory price of $2 per million input tokens and $10 per million output tokens. Google's developer post says to use the paid tier of the Gemini API in Google AI Studio and Android Studio; Ultra subscribers can try it in AI Studio and Antigravity. The earlier limited rollout to trusted cyber defenders began on September 30.
Correction: Updated on 13 Oct: Google says Gemini 4 Argon is now available to developers, enterprises and Google AI Pro and Ultra subscribers, after the limited rollout to trusted cyber defenders that began on September 30.
Who can use it
Developers and enterprises on the paid tier of the Gemini API (gemini-4-argon); Google AI Ultra subscribers in Google Antigravity and AI Studio; Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search. Free-tier access: not specified by source.
When
Available from October 13, 2026. Pricing is introductory; the post does not give a date for the introductory period to end.
Go, Plus and Pro subscribers can let participating apps and sites use their ChatGPT plan allowance without creating or sharing an OpenAI API key. You choose which apps can use it and set a weekly limit per app. Usage counts toward your existing ChatGPT Work and Codex allowance; connecting an app does not add usage. Using purchased credits needs a separate opt-in, and partners may still charge for their own services.
Who can use it
Release notes say eligible Plus and Pro users; OpenAI's help article also lists Go. Only participating apps that support the feature. Release notes describe a limited preview.
When
Listed in the ChatGPT release notes on September 29, 2026.
OpenAI introduced Pro 500, a $500/month plan with the highest included usage of the Pro plans and the only Pro plan with GPT-6 Astra Ultrafast in ChatGPT Work and Codex. Ultrafast uses included allowance at 8 times the Standard rate, then credits. Pro 200 is open to new subscriptions again at $200/month, but new subscriptions that aren't grandfathered get a lower included allowance than Pro 200 previously offered. OpenAI has not published allowance numbers.
Who can use it
Pro 500: new subscriptions through ChatGPT on the web at launch. Buying credits on Pro 100 or Pro 200 does not unlock Ultrafast. Pro 200: grandfathering covers anyone with an active Pro 200 subscription from September 22 to 10 a.m. PT on September 29, 2026.
When
Announced September 29, 2026. Grandfathered Pro 200 subscribers keep the previous allowance through October 29, 2026, then move to the lower allowance. Affected users get an email.
Meta announced Muse for Small Business, adding skills and connectors so its Muse personal AI agent can help run a business. Users can connect tools including Asana, Box, Canva, Dropbox, Figma, Shopify, Slack, Stripe and Zoom, plus Facebook and Instagram business accounts. Meta says nothing publishes, sends or spends without your approval. Custom connectors are also supported.
Who can use it
Muse users. Meta describes Muse as available in the US and Canada, free for most uses with subscription plans for more. Whether the new connectors differ by plan or country: not specified by source.
When
Announced September 29, 2026. Meta says more connectors and skills are coming; it gives no further dates.
OpenAI introduced ChatGPT Space, which brings your Pages, files and related work together in ChatGPT. You create a space for a project, add the material you need and share it with other people so you can keep building on the same work. Space replaces Library for accounts with access; existing Projects remain separate. Pages are new collaborative documents: turn a conversation into a Page, start from a template or begin writing directly in Space, then edit the text yourself or ask ChatGPT to revise it, add charts or create interactive content. Collaborators can work on the same Page and leave comments, each using their own ChatGPT. Pages follow Space's availability and sharing settings, and sharing a Page does not give other people access to your private chats or memory.
Who can use it
Eligible Pro, Business and Enterprise users in the ChatGPT desktop app and on the web. Sharing in managed workspaces follows your organization's settings; Enterprise sharing is an opt-in preview.
When
Listed in the ChatGPT release notes on September 29, 2026. Availability is described as 'eligible' users; no rollout completion date is given.
OpenAI added Ultrafast mode for GPT-6 Astra in the Responses API. Developers use the gpt-6-astra model with service_tier set to "ultrafast" to reduce the time between generated output tokens. OpenAI links separate Ultrafast pricing, which this entry does not restate.
Who can use it
API customers, subject to rate limits. Processing is global with US data residency; EU and other regional inference residency are not supported.
When
Listed in the OpenAI API changelog on September 29, 2026.
Codex Cloud lets you start and continue coding tasks from desktop, web or mobile, using reusable environments with your project's repositories, tools and dependencies. Each task gets its own isolated workspace, and tasks keep running while your computer is asleep. In managed workspaces, the workspace's Codex cloud-access settings apply.
Who can use it
Eligible Plus, all Pro tiers, Business, Enterprise, Healthcare and Education accounts, subject to rollout and workspace settings. Not included with Free or Go. Guest, K-12 and Enterprise view-only seats can't create cloud environments.
When
Listed in the ChatGPT release notes on September 29, 2026. Rollout may vary by account.
OpenAI releases GPT-6.1 Sol, a model update for coding, document work and tasks across apps. OpenAI reports better results than GPT-6 Sol in its tests. Standard API input and output prices are unchanged; reusing cached input costs less.
Correction: Work and Codex access starts with Pro and expands gradually to other eligible plans. Workspace permissions apply. Eligibility does not guarantee access on your account.
Who can use it
ChatGPT Work and Codex: rollout starts with Pro, expanding to Plus, Business, Enterprise and Edu; workspace permissions apply. Also available through the OpenAI API. Not yet in Chat.
When
Announced September 29, 2026. Release notes describe a staged rollout, so access may vary by account. The launch article confirms API availability and says Ultrafast is coming later without a confirmed date.
SpaceXAI launched Team Bots: a Grok Bot that an owner sets up once, with plugins, credentials, skills and files, and publishes to the whole team. Teammates use it from the desktop app, the mobile app and Slack, where each Team Bot has its own handle. Each person's conversations stay private, while shared learnings go to team memory. Usage counts against each teammate's own Grok Bot allowance.
Who can use it
Public beta on Teams and Enterprise plans. Create and publish in the Grok Bot desktop app. Admins choose who gets a Team Bot by default; on Enterprise, audit logs record Team Bot events. Slack may need admin approval.
Anthropic releases Sonnet 5.5 for everyday tasks, coding and creating documents, slides and spreadsheets. Anthropic reports faster responses and lower cost per task than Sonnet 5 in its tests; the underlying API rates are unchanged.
Correction: Sonnet 5.5 handles coding and document work. Anthropic's tests report faster responses and lower cost per task than Sonnet 5. API rates are unchanged.
Who can use it
Claude apps and the Claude API (claude-sonnet-5-5), plus Amazon Bedrock, Claude Platform on AWS, Google Cloud and Microsoft Foundry.
Google adds animated video characters to Gemini voice conversations. Businesses can use them for customer service or guided walkthroughs while the assistant checks information in the background.
Correction: Live Avatar adds animated characters to Gemini voice conversations for customer service or walkthroughs. Gemini Enterprise access is required. Custom avatars need approval.
Who can use it
Available through Gemini Enterprise. Businesses need approval to create custom avatars.
When
Announced September 24; the source says enterprise availability starts that day.
Qwen Cloud lists decision-model-preview, a structured decision model for high-frequency business decisions such as ticket routing, content moderation, agent routing and result validation. It runs classification, yes-or-no decisions and scoring in parallel, returning probability distributions and confidence scores rather than generated text. Context is 65,536 tokens. It does not support streaming, function calling, web search or structured output.
Who can use it
Developers using Qwen Cloud (model name decision-model-preview). The API reference lists a Singapore endpoint. Qwen's model page lists input as free per 1M tokens; output pricing is not specified.
When
Listed in the Qwen Cloud model changelog on September 24, 2026. A preview; no end date given.
Gemini can work with more connected apps, including Linear, Adobe and Peloton. You can manage projects, create designs or plan workouts from the conversation after connecting a supported app.
Correction: Connect supported apps in Gemini settings to manage projects, create designs or plan workouts from a conversation. Access can vary during rollout.
Who can use it
Connect supported apps in Gemini settings. The announcement does not specify a complete plan or region list.
When
Rollout begins September 23; individual availability can vary during rollout.
Google's new Flash TTS and Flash-Lite TTS models turn written scripts into spoken audio. Creators can shape voices, pacing and emotion for podcasts, audiobooks and other projects. Copying a real voice requires the owner's consent.
Correction: Turns scripts into speech. Copying voices requires the owner's consent and excludes Illinois, Texas, EEA, UK, Switzerland and India in AI Studio. Enterprise API access is still planned.
Who can use it
Gemini API and AI Studio; Flash TTS also in Gemini Notebook and Flash-Lite TTS in Google Vids. Enterprise API access is coming soon.
When
Rollout began September 23, 2026. Voice replication in AI Studio excludes Illinois, Texas, the EEA, UK, Switzerland and India; the enterprise API is described as coming soon.
Anthropic launched Claude Marketplace, bringing plugins and connectors, agents and products, and service partners into one place, with more than 2,000 connectors and plugins. Customers can use a portion of their committed Anthropic spend on Claude-powered software from partners.
OpenAI's Codex lead announced a one-time usage reset for Plus, Pro and Business users to save for later. Check Settings > Usage to use it and see its expiry. The reset restores usage once; it does not increase your plan's ongoing allowance.
Correction: This offer was a one-time reset for Plus, Pro and Business users. Check Settings > Usage for any unused reset and its expiry.
Who can use it
Plus, Pro and Business accounts.
When
Announced September 22, 2026. Check your account for availability and the expiry of your reset.
OpenAI introduces GPT-6 Sol and Luna for everyday work and coding. It says API prices for Sol and Luna are 50% lower than their GPT-5.6 promotional pricing. For Sol that is $2 per million input tokens (was $4) and $10 per million output tokens (was $20).
Correction: Updated on 4 Oct: OpenAI says Sol and Luna API prices are 50% lower than GPT-5.6 promotional pricing. Sol is now $2 per million input tokens (was $4) and $10 per million output tokens (was $20).
Who can use it
ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu. Free and Go can use Luna in the desktop app. Both models are in the API; not yet in Chat.
When
Announced September 22. ChatGPT rollout is gradual during the day; access may appear later.
With the Claude Opus 5.5 launch, Anthropic said it is increasing five-hour usage limits on Pro, Max, Team and seat-based Enterprise plans. Its announcement did not state the size. A ClaudeDevs thread the same day gives figures for Claude Code only: 5-hour session limits increase 20%, and because Opus 5.5 is priced lower it goes 25% further within limits.
Correction: Updated on 4 Oct: ClaudeDevs says Claude Code's 5-hour session limits rose 20% on 22 Sep. Anthropic gave no size for the increase on other plans or products in the sources checked.
Who can use it
Pro, Max, Team and seat-based Enterprise plans (the Claude X post lists Pro, Max and Team).
When
Announced 2026-09-22; no separate start or end date given.
Anthropic gave subscribers a one-time reset they can save for later. Redeem it in Settings > Usage on the web or in Claude Desktop to restore the allowance shown in the offer. The reset also applies to shared account usage in mobile and Claude Code. If you downgrade or cancel before using it, it is no longer available.
Correction: Updated on 4 Oct: the reset is no longer available if you downgrade or cancel before using it. Pro, Max and Team users redeem in Settings > Usage on web or Claude Desktop. Use by Oct 22, 2026.
Who can use it
ClaudeDevs names Pro, Max and Team users. Redeem on the web or in Claude Desktop; the reset button is unavailable in mobile and Claude Code, but the restored shared allowance applies there too.
When
Available from September 22, 2026. ClaudeDevs says to use it by October 22; check Settings > Usage for the exact expiry on your offer. It is lost if you downgrade or cancel before using it.
Mistral is merging Chat and Work into one Vibe experience, with one conversation list, Fast and Think modes, Skills replacing Chat Agents and Chat Memories moving to the Knowledge Base. Three Vibe Work features were listed the same day. Spreadsheets in Vibe Work lets you analyze or create Excel and CSV files, edit them in Canvas and download an .xlsx file. Mini App Canvas builds interactive apps and dashboards from a prompt that you can edit and share. Agentic Knowledge Base builds and maintains a browsable personal knowledge base in public preview.
Correction: Updated on 4 Oct: Mistral's release notes list three more Vibe Work features from 22 Sep: Spreadsheets (paid plans, Serverless), Mini App Canvas (Serverless) and Agentic Knowledge Base (public preview). Summary, who gets it and timing now cover them.
Who can use it
Simpler Vibe: Free, Pro, Teams from 2026-09-14; Enterprise from 2026-10-01. Spreadsheets: paid plans, Serverless. Mini App Canvas: Serverless. Agentic Knowledge Base: public preview.
When
Phased rollout started 2026-09-14 for Free, Pro and Teams; Enterprise migration starts 2026-10-01. Spreadsheets, Mini App Canvas and Agentic Knowledge Base were listed on 2026-09-22.
SpaceXAI releases Grok 4.7 for coding, documents and other work tasks. The company says it handles longer jobs and checks its work more carefully than Grok 4.6. Developers can use it in coding tools or build it into their own services.
Correction: Grok 4.7 helps with coding and documents. SpaceXAI says it handles longer jobs and checks its work more carefully than Grok 4.6.
Who can use it
Cursor, Grok Build, the xAI API, third-party coding harnesses, model routers and cloud platforms. Grok 4.7 Fast is only in Cursor and Grok Build.
Qwen Cloud listed three real-time models. Qwen3.8-LiveTranslate-Flash-Realtime translates audio and video live, recognizes 60 source languages, speaks in 29 and cuts latency to about 2.3 seconds. Qwen-Audio-3.1-Realtime-Plus handles live speech conversations, adding eight system voices, function calling, web search and voice cloning. Qwen3.8-Omni-Flash-Realtime handles real-time audio and video interaction and adds multichannel audio and remote MCP tools.
Who can use it
Developers using Qwen Cloud's real-time APIs. Plans, regions and pricing: not specified by source.
When
Listed in the Qwen Cloud model changelog on September 17 (LiveTranslate), 20 (Audio-3.1-Realtime-Plus) and 21 (Omni-Flash-Realtime), 2026.
Qwen Cloud lists qwen3.8-omni-flash, a next-generation native omni-modal model. It supports context lengths of up to 1M tokens and accepts text, image, audio and video inputs. Qwen says it is designed for agentic work such as coding, knowledge work and GUI interaction, and it supports two-channel and four-channel spatial audio understanding. It works with the DashScope and OpenAI protocols.
Who can use it
Developers using Qwen Cloud (model name qwen3.8-omni-flash). Plans, regions and pricing: not specified by source.
When
Listed in the Qwen Cloud model changelog on September 18, 2026.
Claude Code projects can split a larger job into several tasks, coordinate the work and bring the results together. You can guide progress in one project conversation. Running several tasks at once can use your allowance faster.
Correction: Initial beta was limited to selected Pro and Max users of Claude Code cloud sessions with no existing web or desktop projects. The source does not confirm completion of the wider rollout.
Who can use it
Initial beta: selected Pro and Max users of Claude Code cloud sessions with no existing web or desktop projects. Broader Pro/Max access comes later, followed by Team and Enterprise; a waitlist is available.
When
Initial beta announced September 17, 2026. The article planned wider Pro/Max access over the following week; it does not confirm completion of the rollout.
xAI made grok-voice-transcribe-2.0, a new speech-to-text model, available on its API alongside grok-voice-transcribe-1.0, with 2.0 as the default. xAI later ended grok-voice-transcribe-1.0 on October 2, 2026: requests to that model name are now routed to 2.0 at the same price, with higher accuracy.
Correction: Updated on 4 Oct: xAI ended grok-voice-transcribe-1.0 on 2 Oct. Requests to the 1.0 model name are now routed to 2.0 at the same price, with higher accuracy.
Who can use it
xAI API users.
When
2.0 available from 2026-09-17. grok-voice-transcribe-1.0 reached end of life on 2026-10-02.
ChatGPT is now available as a sidebar add-in in Microsoft Word, to draft from notes, summarise documents, revise selected text and adjust formatting, joining the Excel and PowerPoint add-ins.
Correction: Updated on 4 Oct: OpenAI says ChatGPT for Word is enabled by default from 1 Oct 2026. Workspace admins can still turn it off, and Microsoft 365 admins must allow the add-in for it to appear in Word.
Who can use it
All ChatGPT plans, including Free. Workspace admins can turn it on or off, and Microsoft 365 admins must also allow the add-in. Usage can reduce the shared allowance for Codex and other premium features.
When
Available from 2026-09-17. OpenAI's help page says that starting October 1, 2026 ChatGPT for Word will be enabled by default.
Google launched Guided Vision in Gemini Live on October 1, 2026. It was built with and for blind and low-vision communities. When you share your camera, Gemini describes what is in front of you aloud and gives spoken prompts to help you reframe. Google says it is an assistive tool that can make mistakes, and is not for navigation, safe travel or obstacle detection.
Correction: Updated on 4 Oct: Google launched Guided Vision in Gemini Live on 1 Oct, after first calling it coming soon on 1 Sep. It works on Android 9 and above in regions and languages where Gemini Live is supported.
Who can use it
Android 9 and above, in regions and languages where Gemini Live is supported. Turn on 'Use Guided Vision in Live' in the Gemini app, or use an Accessibility shortcut or TalkBack.
When
Previewed September 1, 2026 as 'coming soon'; launched October 1, 2026 across compatible Android devices.
OpenAI shut down a set of older API models on October 23, 2026, after announcing their deprecation on April 22, 2026. The list includes gpt-3.5-turbo (gpt-3.5-turbo-0125), gpt-4 (gpt-4-0613), gpt-4-1106-preview, gpt-4-turbo, gpt-4.1-nano, gpt-4o-2024-05-13, gpt-image-1, o1, o1-pro, o3-mini and o4-mini, plus fine-tuned versions of gpt-3.5-turbo, gpt-4, gpt-4.1-nano, babbage-002 and davinci-002, and ft-o4-mini-2025-04-16. Recommended substitutes are gpt-5.6-terra, gpt-5.6-sol (with reasoning.mode: pro for o1-pro), gpt-5.6-luna and, for gpt-image-1, gpt-image-2.5-sunburst or gpt-image-2.5-flare. OpenAI says some developers may be able to provision dedicated capacity for continued access after a shutdown date.
Correction: Corrected 22 Oct 2026: we added the fine-tuned models ft-davinci-002 and ft-o4-mini-2025-04-16, fixed the substitute for gpt-image-1 to gpt-image-2.5-sunburst or gpt-image-2.5-flare, and removed a claim that ChatGPT plans are unaffected, which OpenAI's page doesn't make.
Who can use it
OpenAI API developers still calling any of the listed snapshots or their fine-tuned versions.
When
Deprecation announced April 22, 2026. OpenAI's deprecations page now says access 'was shut down on October 23, 2026'; no time was given.
OpenAI and Google Cloud are partnering to bring OpenAI's frontier models and Codex agentic workflows into Google Cloud's Model Garden. Google Cloud customers will be able to buy the models through Model Garden, transacted through Google Cloud Marketplace, and enterprises with Google Cloud spend commitments can draw them down against those purchases. OpenAI models will soon also be available in Google Antigravity and the Gemini Agent. OpenAI says Microsoft remains its primary cloud partner. Pricing and the list of models were not published.
Who can use it
Google Cloud customers, purchased through Model Garden and Google Cloud Marketplace. Access is by request for now.
When
Announced October 21, 2026. General availability is planned in the coming weeks; no date was given.
DeepSeek says its Search and Fetch APIs, powered by its in-house search engine, are now open to everyone. Each endpoint costs $0.75 per 1,000 calls. Search and Fetch calls made through server-side tools in DeepSeek Harness, Claude Code and Codex are free for DeepSeek API users until billing starts.
Who can use it
Open to everyone through the DeepSeek API. The free period covers server-side tool calls in DeepSeek Harness, Claude Code and Codex for DeepSeek API users.
When
Announced October 17, 2026. Billing starts October 31, 2026 at 16:00 UTC.
OpenAI notified developers on October 14, 2026 that gpt-4o (2024-08-06 and 2024-11-20), gpt-4o-mini (2024-07-18), gpt-4.1 (2025-04-14) and gpt-4.1-mini (2025-04-14) are deprecated and will be removed from the API on January 14, 2027. OpenAI recommends gpt-6-luna as the replacement for all four. Fine-tuned versions follow the same shutdown schedule as their base models.
Who can use it
OpenAI API developers using these GPT-4o and GPT-4.1 snapshots, including fine-tuned versions of them.
When
Deprecation notified October 14, 2026. Removal January 14, 2027, given as a calendar date with no time.
Google deprecated the Deep Research agent deep-research-pro-preview-12-2025 in the Gemini API and will shut it down on October 23, 2026. Developers should move to deep-research-preview-04-2026, which Google describes as designed for speed and efficiency and ideal to be streamed back to a client UI, or deep-research-max-preview-04-2026, described as maximum comprehensiveness for automated context gathering and synthesis. To migrate, change the agent parameter in interactions.create requests.
Who can use it
Gemini API developers using the deep-research-pro-preview-12-2025 agent.
When
Deprecation announced October 8, 2026. Shutdown October 23, 2026.
OpenAI says it will soon add College Planner to ChatGPT for Teens. It brings together application requirements, deadlines, tasks and financial-aid steps for the schools on a student's list in one plan. The first version is being designed for US high-school students in grades 10–12 who plan to attend a four-year college. OpenAI is also adding flashcards to ChatGPT for Teens. No date was given for College Planner.
Who can use it
ChatGPT for Teens. College Planner is initially for US students in grades 10–12 planning to attend a four-year college.
When
Announced October 7, 2026. College Planner is coming 'soon', with no date given.
Google deprecated the gemini-3.1-flash-image model, also known as Nano Banana 2, in the Gemini API. Google has not announced a shutdown date. Google tells developers to migrate to gemini-nano-banana-2.1, released the same day, which it describes as an update that maintains Flash-level speed and cost efficiency. The deprecations page lists gemini-3.1-flash-image with a release date of May 28, 2026, no shutdown date announced, and gemini-nano-banana-2.1 as the recommended replacement.
Correction: Corrected 22 Oct 2026: Google has not announced a shutdown date for gemini-3.1-flash-image. An earlier version of this page said 29 October 2026, which was wrong.
Who can use it
Gemini API developers using gemini-3.1-flash-image. The replacement gemini-nano-banana-2.1 has no shutdown date announced.
When
Deprecation announced October 6, 2026. No shutdown date announced.
OpenAI introduced a new visual ad format for ChatGPT that shows images of product inspiration, product usage or the experiences they make possible. Initially OpenAI will test the format during image generation in ChatGPT for users on the Free and Go plans. OpenAI says ads will be clearly labeled and remain separate from the image being created, and that advertising does not influence the answers ChatGPT provides. Testing will begin later this month in the US with an initial group of advertisers. The same post expands measurement partners for ChatGPT Ads and makes Negative Phrases available to qualifying advertisers.
Who can use it
Free and Go plan users in the US during the initial test. Advertisers: an initial group at first; businesses can sign up at ads.openai.com.
When
Announced October 5, 2026. OpenAI says testing will begin later this month (October 2026) in the US and gives no exact date.
Meta says it will bring Muse, its personal AI agent, to its AI glasses for hands-free use, activated by saying the agent's name. Muse can also act on what you are looking at. The post is from Meta's India newsroom, which also says Ray-Ban Meta (Gen 3) is now on sale in India from INR 44,300 and that Ray-Ban Meta Audio glasses are coming soon.
Who can use it
Muse on glasses: not specified by source. Ray-Ban Meta (Gen 3): on sale in India at meta.com, in.rayban.com and retailers. Ray-Ban Meta Audio: coming soon.
When
Announced October 1, 2026. Muse on AI glasses is 'in the coming months'; no date or country list is given.
OpenAI deprecated gpt-5.3-codex, gpt-5.4-nano and gpt-5.1. They will be removed from the API on April 1, 2027, with six months' notice. OpenAI recommends migrating gpt-5.3-codex and gpt-5.1 to gpt-6-sol, and gpt-5.4-nano to gpt-6-luna, before the shutdown date.
Who can use it
OpenAI API developers using gpt-5.3-codex, gpt-5.1 or gpt-5.4-nano.
When
Deprecation announced October 1, 2026. Removal April 1, 2027, given as a calendar date with no time.
Anthropic announced the deprecation of Claude Sonnet 4.5 (claude-sonnet-4-5-20250929). It will be retired on the Claude API on November 30, 2026. Anthropic recommends migrating to Claude Sonnet 5.5 (claude-sonnet-5-5). Anthropic's deprecations page says it notified developers using Sonnet 4.5 on September 30.
Correction: Corrected 22 Oct 2026: added the platforms Anthropic names. The date applies to the Claude API, Claude Platform on AWS and Microsoft Foundry; Amazon Bedrock and Google Cloud set their own dates. An earlier version said the source didn't specify them.
Who can use it
Developers calling claude-sonnet-4-5-20250929 on the Claude API. Applies to the Claude API, Claude Platform on AWS and Microsoft Foundry. Amazon Bedrock and Google Cloud set their own dates.
When
Announced September 30, 2026. Retirement is scheduled for November 30, 2026, a calendar date rather than an exact time.
OpenAI introduces dot, an agent in ChatGPT that can continue working between conversations. It uses GPT-6 Astra and a cloud computer to work across connected apps, retain context and bring results back for review. Users define its goals and what it may do independently.
Correction: Updated: OpenAI's Help Center now says you can create your dot in the ChatGPT mobile app on iOS or Android, as well as in the desktop app or on desktop web. Dots are still not supported on mobile web. The earlier text said mobile creation was not supported; the September 29 release note named only the desktop app and desktop web.
Who can use it
Adults 18+: Pro outside the EEA, Switzerland and UK at launch; Business Premium in supported ChatGPT regions. Enterprise, Edu and Healthcare beta requires an admin to enable it. Create in the iOS or Android app, the desktop app or on desktop web; access is gradual.
When
Announced September 29, 2026. Rollout may take several days per account. No exact release hour or completion date is confirmed. Dots are not supported on mobile web.
Meta announced Meta Enterprise Platform, a new business pillar to help companies use AI. Initially it will bring Meta's technology stack, including the Muse agent, Meta Business Agent, Muse API and Muse Code, to businesses and developers. Chirantan 'CJ' Desai, previously CEO of MongoDB, joins Meta as Chief Enterprise Platform Officer, reporting to Mark Zuckerberg.
Who can use it
Not specified by source.
When
Announced September 28, 2026. The post gives no product availability dates.
OpenAI plans to retire custom GPTs and help creators move them to plugins. Existing GPTs remain usable until their applicable retirement date. Test the replacement before switching: custom actions and sharing settings do not transfer automatically.
Correction: Updated: OpenAI's FAQ now says new GPT creation ended on October 26, 2026 for affected Enterprise workspaces, with the exception of FedRAMP workspaces, so the timing no longer calls it planned. It also names FedRAMP workspaces, along with affected Enterprise workspaces with a qualifying, approved deferral, for the February 11, 2027 retirement date. Retirement is still December 11, 2026.
Who can use it
All ChatGPT plans; migration and plugin access vary by account or workspace. FedRAMP workspaces and affected Enterprise workspaces with a qualifying, approved deferral have a later retirement date; it does not apply to every workspace. Follow your account notice.
When
Retirement December 11, 2026; February 11, 2027 for FedRAMP workspaces and affected Enterprise workspaces with a qualifying, approved deferral. OpenAI says new GPT creation ended October 26, 2026 for affected Enterprise workspaces other than FedRAMP. Calendar dates, not exact times.
xAI will retire the grok-imagine-image-quality model on its API on November 2, 2026. The model name keeps working: requests will be served by grok-imagine-image-2.0 at low quality, with the same request and response shape and $0.01 less per image. xAI recommends switching to grok-imagine-image-2.0 before then so you can choose the quality you pay for. grok-imagine-image-pro already redirects here and will follow it.
Who can use it
Developers calling grok-imagine-image-quality or grok-imagine-image-pro on the xAI API. grok-imagine-image is not affected.
When
Notice given September 2, 2026. Retirement November 2, 2026, a calendar date rather than an exact time.
On June 11, 2026, OpenAI notified developers using older GPT-5 and o3 model snapshots of their deprecation and removal from the API on December 11, 2026. The snapshots are gpt-5-2025-08-07, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-5-pro-2025-10-06, o3-2025-04-16 and o3-pro-2025-06-10. Recommended replacements are gpt-5.6-sol (with reasoning.mode: pro for the pro snapshots), gpt-5.6-terra for gpt-5-mini and gpt-5.6-luna for gpt-5-nano.
Who can use it
OpenAI API developers pinned to the listed 2025 GPT-5 and o3 snapshots.
When
Deprecation announced June 11, 2026. Shutdown December 11, 2026, given as a calendar date with no time.
On June 3, 2026, OpenAI notified developers using Agent Builder that the product is being deprecated. Agent Builder is scheduled to shut down on November 30, 2026. ChatKit remains available. OpenAI points developers to its Migrate from Agent Builder guide to continue with the Agents SDK or ChatGPT Workspace Agents.
Who can use it
Developers with agents built in OpenAI's Agent Builder. ChatKit users are not affected.
When
Deprecation announced June 3, 2026. Shutdown November 30, 2026, given as a calendar date with no time.
On June 2, 2026, OpenAI notified developers using older GPT Image models of their deprecation and removal from the API on December 1, 2026. The models are gpt-image-1-mini, gpt-image-1.5 and the chatgpt-image-latest alias. OpenAI recommends gpt-image-2.5-sunburst or gpt-image-2.5-flare as replacements. The earlier gpt-image-1 model is on a separate schedule and shuts down on October 23, 2026.
Who can use it
OpenAI API developers using gpt-image-1-mini, gpt-image-1.5 or chatgpt-image-latest. Image generation inside ChatGPT is not affected by this API notice.
When
Deprecation announced June 2, 2026. Removal December 1, 2026, given as a calendar date with no time.
Gemini
ChatGPT
Claude
Codex
Mistral
DeepSeek
Grok
Muse models
Qwen
Muse
Grok Bot
dot