What changed, what the official sources say, and what it means for your workflow.
Published by SessionWatcher. AI assisted reporting with a separate factual review. Read our editorial process.
As of October 7, 2026, GitHub Copilot's official supported-models page lists Claude Haiku 5.5 as generally available. The supplied page does not establish its launch date or which specific plans and clients can access it.
As of October 7, Cursor's pricing page lists Claude Haiku 5.5 at $0.10 per million input tokens and $0.50 per million output tokens. Claude Sonnet 5.5 cache reads fall from $0.20 to $0.10 per million tokens. The page does not specify an effective date.
OpenAI reported Dot turn failures, delayed responses and reduced proactivity in Codex and Work mode on October 7, 2026. The status-page timeline runs from 5:21 PM to full recovery at 6:23 PM, a 62-minute incident.
OpenAI reported that Dot in Codex could fail to create new threads on October 7, 2026. The status-page timeline runs from 4:31 PM to full recovery at 6:06 PM, a 95-minute incident.
OpenAI resolved the outage blocking new ChatGPT Work threads at 11:53 AM on October 7. Codex in ChatGPT Desktop was listed as affected. Fixed app versions are available for Linux, macOS and Windows; affected users should update.
Anthropic's October 7 release notes add Claude Haiku 5.5 as the default Haiku model on its API. Listed pricing is $0.10/$0.50 per million tokens, rising to $0.50/$2.50 for prompts over 100K tokens.
Anthropic's October 7 announcement halves Claude Sonnet 5.5 prompt-cache read prices from $0.20 to $0.10 USD per million tokens. Cache writes and all other prices are unchanged.
OpenAI reported elevated error rates in ChatGPT Go conversations on October 6, 2026. Investigation began at 1:35 PM, mitigation was applied by 1:53 PM, and full recovery was confirmed at 2:40 PM.
OpenAI's updated announcement says Auto-Review is now free, marking Day 2 of its Codex and Work improvement campaign. It does not specify eligible plans or how reviews count toward usage limits.
OpenAI's October 2 release notes for ChatGPT for iOS add Home Screen and Lock Screen widgets showing remaining usage and reset times. Home Screen widgets also show recent tasks from selected computers.
OpenAI's status page reports degraded performance on October 6, 2026, affecting ChatGPT conversations, ChatGPT Work and other services, including API rehydration. The incident ran from 02:41 AM to 03:17 AM, when OpenAI reported full recovery.
Z.ai's official GLM Coding Plan page now states that all-day usage consumes credits at the off-peak rate from September 25 to October 7, 2026. It also lists an overnight GLM-5.3-Flash campaign with unlimited usage in ZCode and AutoClaw for paid users.
Claude Code v2.1.290 changes the interactive session's WebSearch budget from a fixed 200-call cap to a budget that refills at 100 calls per hour. An environment variable sets the refill rate, and setting it to 0 turns refilling off.
Claude's status page reports elevated errors for Claude Mythos 5.1 and Claude Fable 5.1 on October 5, 2026. Impact lasted from 12:40 to 13:10 UTC and affected Claude Code, the Claude API, claude.ai and Claude Cowork.
OpenAI says that over 28 days it will deliver a clear improvement relevant to most Codex and Work users each day, or a full reset instead. The announcement does not define which usage limits a full reset would cover.
OpenAI reports recovery from elevated Work Mode errors affecting Codex in ChatGPT Desktop and ChatGPT Work. Investigation began October 5 at 07:03 AM, with mitigation at 09:50 AM and a resolved update October 6 at 02:10 AM.
Google's Antigravity models page now says Claude Sonnet 4.6, Claude Opus 4.6 and GPT-OSS-120b will be removed on November 2, 2026. The listed models currently have different availability across Google AI plans.
Google's support page says Gemini Apps model availability will change for personal accounts starting in October 2026. Changes begin October 9 for users without an AI subscription; AI Plus subscribers will receive their effective date by email.
Claude Code v2.1.278 defaults auto mode to a server-side classifier for Claude API and Enterprise users, plus named cloud platforms and gateways. The classifier does not charge for classifier overhead, and /status now shows whether it runs on the server.
GitHub reports that approximately 70% of requested Copilot Code Reviews did not complete during a September 28 disruption from 19:23 to 22:08 UTC. Failed reviews were not automatically retried, so users needed to request them again.
Google's September 18 update limits Gemini 2.5 model access to users who actively used them previously. The models are not deprecated and remain served through the API until further notice; Google recommends 3.5 Flash-Lite or 3.8 Flash for new projects.
Claude's status page reports an outage from 14:00 to 14:59 UTC on September 29, affecting Claude Code, the API and other Claude services. Sessions and sign-ins recovered, but some messages sent during the incident may not have been saved.
Purchased credits were delayed in Claude Platform account balances on October 1, causing some requests to fail. Anthropic reported the issue at 16:20 UTC, reported a fix at 18:23 UTC, and marked the incident resolved at 22:36 UTC.
Cursor reports that changes to its agent harness reduced token costs for users by 7% without reducing agent quality. The changes include shorter system prompts, loading tools only when needed and improved cache reuse.
OpenAI reported higher-than-expected charges for hosted containers in the Agent API on September 18 and marked the incident resolved September 19. It said it was reviewing affected usage to identify customers and calculate refunds.
OpenAI's DevDay recap says Pro 200 subscriptions are open again, with different allowances for new subscriptions than grandfathered plans. It also reports a worldwide usage reset, without specifying its timing or scope.
Anthropic announced the deprecation of Claude Sonnet 4.5 on September 30, 2026. Retirement on the Claude API is scheduled for November 30, 2026, and Anthropic recommends migrating to Claude Sonnet 5.5.
OpenAI says GPT-5.5 will retire from ChatGPT, ChatGPT Work and Codex on October 14, 2026, across all plans, but not from the API. GPT-5.3-Codex-Spark was deprecated September 14 and is no longer available in the listed clients.
OpenAI's GPT-6 caching announcement highlights explicit breakpoints and diagnostics. Companion API documentation lists cache writes at 1.25 times uncached input pricing for GPT-5.6 and later, with discounted reads and controls to avoid caching changing content.
Devin says users can sign in with ChatGPT and use an eligible ChatGPT Plus or Pro plan for GPT model usage in Devin Desktop. GPT requests are billed to the ChatGPT plan; the release note links to setup instructions for eligible models.
Kimi says cached content for kimi-k3 is billed at one-tenth the cache-miss price. Cache Write is billed separately, with 5-minute and 1-hour TTL options; omitting the TTL setting defaults to 5 minutes.
GitHub says Gemini 3.7 Flash, GPT-5.5, GPT-5.4, GPT-5.4 mini, GPT-5 mini and Grok 4.5 will be deprecated across Copilot experiences on October 19, 2026. The announcement lists suggested replacement models.
GitHub deprecated Gemini 3.5 Flash, Gemini 3.6 Flash, Kimi K2.7 Code and Claude Opus 4.7 across Copilot experiences on October 2, 2026. It suggests Gemini 3.8 Flash, Kimi K3 and Claude Opus 5.5 as alternatives.
GitHub Copilot auto now offers efficiency, balance and intelligence tiers to guide how it weighs cost, quality and response time. All tiers use the same available models, and usage is billed according to the model auto selects.
GitHub says Copilot Business and Enterprise users with usage-based billing can request a higher budget after reaching their AI credit limit. Organization or enterprise owners and billing managers can approve, adjust, or deny requests; approval restores access to AI credits.
GitHub says Claude Opus 5.5 and GPT-6 Sol are available to Copilot Pro+, Max, Business and Enterprise plans. GPT-6 Luna and Grok 4.7 are also available, with Pro added to their listed plans. The Copilot app also adds local sandboxing in public preview.
Claude Code v2.1.280 changes the default model on Pro and Team Standard plans from Sonnet to Opus. The release identifies Opus 5.5 as the new default Opus model.
Claude Code v2.1.284 says Sonnet 5.5 is now the default Sonnet model on the Anthropic API. For Claude apps gateway users running this version or later, /usage and the status line can show dollar amounts against the monthly spend limit.
Anthropic announced on September 17 that redesigned Claude Code Projects can coordinate parallel threads, each a full Claude Code cloud session. It warned that projects can reach usage limits faster; the beta rollout was planned to expand in stages.
OpenAI's status page says elevated errors affected Codex, ChatGPT and the API on September 29, 2026. The incident began at 5:52 PM and was resolved at 11:14 PM, with affected services reported recovered.
OpenAI says its Ultrafast speed tier for Astra is available in Codex and ChatGPT Work on Enterprise plans and through the new Pro 500 plan. The company says Codex generation reaches up to 300 tokens per second; Ultrafast is also available to developers through the API.
Anthropic says Opus 5.5 input and output token prices fell 20% for token-billed usage, while cached-token reads fell 60%. Claude Code users can also set effort levels without resetting cache, and API-key and cloud-provider users can set a one-hour cache lifetime.
Google says Gemini 4 Argon will launch at $2 per million input tokens and $10 per million output tokens, with cached input 95% off. After an introductory period, prices rise to $4 and $20 per million tokens respectively. Wider access is planned to start with paid API customers and Google AI Ultra subscribers.
OpenAI says ChatGPT dots can create Work or Codex tasks on a connected computer and continue existing local Codex tasks. Conversations with a dot do not count toward ChatGPT usage limits, but tasks it starts or manages in Work or Codex count toward those products’ limits as usual.
Anthropic says it is resuming billing for refusals that arrive before any output in the bio, frontier_llm, and reasoning_extraction categories. Those refusals are charged at the rate of the model that ran; other pre-output refusal categories remain unbilled, and fallback credit is unchanged.
GitHub has expanded the HydraFusion research preview from Copilot CLI to Visual Studio Code and the GitHub Copilot app. Eligible users can select it in the model picker; organization and enterprise users may need an administrator to enable preview features.
OpenAI says GPT-6 Sol and GPT-6 Luna are rolling out to Plus, Pro, Business, Enterprise, and Edu users in Codex and ChatGPT Work. Free and Go users can access Luna in the desktop app. Availability depends on rollout and workspace settings; Enterprise administrators must enable the models.
GitHub says GPT-6.1 Sol is generally available to Copilot Pro+, Max, Business, and Enterprise users, with gradual rollout across listed Copilot surfaces. Usage-based billing follows provider list pricing; Business and Enterprise administrators can manage access in Copilot settings.
xAI says Grok 4.7 is available in Cursor and Grok Build, as well as through its API and other listed channels. API pricing starts at $2 per million input tokens and $6 per million output tokens; a faster variant costs twice as much.
OpenAI lists GPT-6.1 Sol for Responses and Chat Completions, with standard pricing for prompts up to 272K input tokens of $2 per million input tokens, $0.10 per million cached input tokens, $2.50 per million cache-write tokens, and $10 per million output tokens. Multi-agent support is in beta through a Responses API request.
Anthropic says Claude Sonnet 5.5 is available on several named API and cloud platforms. It also documents compatibility changes affecting thinking settings, forced tool use, thinking blocks, computer use, and advisor-tool requests.
Anthropic lists Opus 5.5 API input and output at $4 and $20 per million tokens, below Opus 5 prices. It also says five-hour usage limits are increasing on Pro, Max, Team, and seat-based Enterprise plans, with a reset subscribers can save and use when they choose.
GitHub says Claude Sonnet 5.5 is generally available to Copilot Pro, Pro+, Max, Business, and Enterprise users, with gradual rollout across listed Copilot surfaces. Business and Enterprise administrators can manage model access in settings.
OpenAI says GPT-6 Sol and GPT-6 Luna accept text and image inputs and generate text through the Responses and Chat Completions APIs. Standard pricing for prompts up to 272K input tokens is $2 input, $0.20 cached input, and $10 output per million tokens for Sol; Luna is $0.10, $0.01, and $0.50 respectively.