AI Tool Watch

Recent AI Tool Changes

This page tracks verified changes to AI tools across ChatGPT, Claude, Gemini, Copilot, Grok, Perplexity, and Meta AI. 30 changes recorded in the last 30 days. Most recent: [Verified] Added the current Voice Mode beta across Free, Pro, Max, Team, and En.

All Recent Changes

2026-10-04

  • Claude Voice Mode [Verified] Added the current Voice Mode beta across Free, Pro, Max, Team, and Enterprise; recorded mobile, desktop, Linux beta, and web surfaces; preserved regular plan usage limits and the explicit Claude Code exclusion Affects: Hear Audio and Speech, Speak Back in Real Time

2026-09-28

  • Grok Grok Image Generation (Aurora) Per x.ai/pricing, image generation is included on every grok.com plan including Free; Gating corrected to free and grok.com tier rows added. Per-tier counts (~10/2hrs, ~20-50/day, ~500/day) removed as unsourced. X Premium rows not confirmed on that page, so Verified is unchanged Affects: Generate Images
  • Grok Grok Imagine (Video Generation) Per x.ai/pricing: video generation not included on Free (was ⚠️), included on SuperGrok Lite and up; SuperGrok Plus adds 1080p and a larger shared usage pool. Unsourced daily counts removed and marked Unverified, resolving the #631 Limits restructure. X Premium rows not confirmed, so Verified is unchanged Affects: Generate Video
  • ChatGPT Chat [Verified] Model lineup corrected: file was still describing GPT-5.4 (Mar 2026) as current, but ChatGPT had since moved to GPT-5.5 and is now transitioning to GPT-6 (Astra rolled out 2026-09-04 to Pro/Enterprise/Business-Premium + API; Sol and Luna rolling out 2026-09-22 to Plus/Pro/Business/Enterprise/Edu and Free/Go respectively). GPT-5.5 is scheduled to retire from ChatGPT/ChatGPT Work/Codex on all plans 2026-10-14. Availability table and talking point updated; exact per-plan GPT-6 rollout status may still be in progress from one verification to the next Affects: Write and Explain, Use It on My Surfaces
  • ChatGPT Codex Security [Verified] Added open-source Codex Security CLI (npm install @openai/codex-security, beta, requires access), launched 2026-07-29; terminal row ❌ → ✅, Linux row ❌ → ⚠️ Affects: Write and Edit Code, Take Actions and Run Tools, Connect to External Systems
  • ChatGPT DALL-E Image Generation [Verified] ChatGPT's image generation moved from GPT Image 1.5 to **ChatGPT Images 2.5** (rolled out 2026-09-08 to all ChatGPT/ChatGPT Work/Codex users); availability notes and talking point updated; noted gpt-image-1 (retires 2026-10-23) and gpt-image-1.5 (retires 2026-12-01) are themselves being superseded by gpt-image-2.5-sunburst/flare per OpenAI's deprecations page Affects: Generate Images
  • ChatGPT Sora (Video Generation) [Verified] API row corrected: the Sora 2 models and Videos API were shut down 2026-09-24 with no one-to-one replacement API; API availability ✅ → ❌ Affects: Generate Video
  • Claude Artifacts [Verified] URL updated (article renumbered to 17153992); added missing Team/Enterprise rows to Availability; added Design/Slides/Docs templates (beta, paid plans, off by default on Enterprise); Claude Code (terminal) corrected from unavailable to available — can publish session output as an artifact or make Design/Docs; talking point rewritten to match Affects: Make and Edit Documents, Write and Edit Code
  • Claude Chat [Verified] Model reference corrected: Sonnet 4.6 is several generations stale (Anthropic shipped Sonnet 5 then Sonnet 5.5, announced today, 2026-09-28); Free-tier default updated to Sonnet 5.5. Fixed Max-tier multiplier mix-up in the Availability table: Max 5x had "20x Pro usage" and Max 20x had "Unlimited standard" — corrected to 5x and 20x respectively per anthropic.com/pricing and the Max plan FAQ ($100/mo = 5x Pro usage, $200/mo = 20x Pro usage). Noted the in-progress Cowork+chat merge. claude.ai itself returned HTTP 403 to automated fetches (bot-blocked); anthropic.com/pricing and the support.claude.com model-settings article served as the readable primary sources Affects: Write and Explain, Use It on My Surfaces
  • Claude Claude Code [Verified] Major platform expansion confirmed via code.claude.com docs: native JetBrains IDE plugin added (talking point previously said none existed); Claude Code on the web (claude.ai/code) added; Claude mobile app (iOS/Android) added for monitoring/steering cloud sessions and Remote Control; Chrome extension (browser automation, paid plans) added. Dropped stale claudelog.com VS Code source (unofficial) Affects: Write and Edit Code, Take Actions and Run Tools, Use It on My Surfaces
  • Claude Connectors [Verified] Old "Setting Up Integrations" source (support.anthropic.com/.../10168395) is dead (301 to support.claude.com, then 404); replaced with current "Use connectors to extend Claude's capabilities" and "custom connectors using remote MCP" articles. Corrected Free-plan notes: custom connectors (remote MCP) are capped at 1 on Free, not unlimited/"full". Mobile (iOS/Android) can now install connectors directly (beta), not desktop-setup-only. Dropped unverifiable "75+ services" and "Deep Connectors" branding not found in current sources Affects: Connect to External Systems
  • Claude Cowork Mode [Verified] Major platform expansion per current "Get started with Claude Cowork" article: web (claude.ai), Claude Mobile (iOS/Android), and Chrome side panel added — no longer desktop-only; Windows arm64/x64 restriction no longer documented, dropped. Added new "Claude Cowork and chat are one Claude" source: Anthropic is merging Cowork and chat into one experience, rolling out gradually to Pro/Max first; talking point and Regional updated to reflect this Affects: Do Multi-Step Research, Take Actions and Run Tools
  • Claude Extended Thinking [Verified] Model lineup and effort levels are far out of date: current flagship models are Sonnet 5.5 / Opus 5.5 / Fable 5.1 (not Opus 4.6), thinking is now mandatory (cannot be disabled) on Sonnet 5.5, Opus 5.5, Fable 5.1, and Opus 5, and a fifth effort level "Extra high" (xhigh) exists between high and max on Opus 4.7 and newer. Talking point rewritten per support.claude.com's model/effort/thinking settings article; added Opus 5.5 and Sonnet 5.5 announcement sources
  • Claude Memory [Verified] Source article renamed/merged: "How does memory work" (support.anthropic.com/.../10166267) now 404s; current article is "Use Claude's chat search and memory to build on previous context," which bundles memory with a new, related but distinct **paid-only** "search past chats" (RAG) capability not available on Free. Memory's own gating (free on Free/Pro/Max, owner-enabled on Team/Enterprise) is unchanged. Talking point updated to distinguish the two features and avoid implying chat search is free Affects: Remember Context Over Time
  • Claude Projects [Verified] URL updated to support.claude.com (support.anthropic.com 301s there); Free plan limit clarified as capped at 5 projects (was listed as unqualified "Standard/Full access"); added "Enhanced project knowledge with RAG" — paid plans (Pro/Max/Team/Enterprise) auto-scale project knowledge up to 10x near context limits; talking point updated; noted (not added as a claim) a new beta "Projects" experience rolling out via Claude Code Affects: Organize Work in Projects, Use Files I Provide
  • Claude Skills [Verified] Gating changed from paid to free — "How to create custom skills" (support.claude.com, dated July 22, 2026) now states plainly "Skills are available for users on Free, Pro, Max, Team, and Enterprise plans," contradicting the prior "custom skills require Pro and above" claim; only the newer "record a skill" video-capture creation flow remains Pro/Max/Team-only (Claude for Mac). Old docs.anthropic.com/en/docs/skills source 404s after redirect; replaced with current docs.claude.com Agent Skills overview and the how-to article Affects: Build Reusable AI Workflows
  • Claude Vision (Image Understanding) [Verified] URL updated — docs.anthropic.com now 301s to platform.claude.com; content re-confirmed, no claims changed Affects: See Images and Screens, Read Text and Documents
  • Gemini Agentic App Actions (Android) [Correction] Replaced secondary source (Deccan Herald) with the official Pixel Phone Help article as primary/URL. Google's own term for this feature is "screen automation," distinct from Gemini Spark and Connected Apps. Device list corrected/expanded to Pixel 10, Pixel 10 Pro, Pixel 10 Pro XL, Samsung Galaxy S26 series, Z Flip 8, and Z Fold 8 (Z Flip 8/Fold 8 also unlock extra categories: buy/sell, travel booking, event tickets). Regional note corrected: Pixel 10 devices excluded from Korea; added 18+/personal-account/English-Korean requirements. Talking point's named third-party apps (Uber, DoorDash, Grubhub, Starbucks, Instacart, Lyft) are not confirmed anywhere in the current official documentation, which describes task categories rather than named partners, so they were removed and replaced with the officially documented task categories Affects: Connect to External Systems, Take Actions and Run Tools
  • Gemini AI Studio [Verified] Re-checked against ai.google.dev/gemini-api/docs/billing, the source that actually substantiates the free-gating claim (the two previously-listed sources are a JS-rendered landing page and a docs page with no pricing text). It confirms "AI Studio usage remains free of charge unless users link a paid API key," and that a linked billing account moves projects to higher rate limits/advanced models — matching the existing Free/Google AI Pro rows and talking point. No claim changes Affects: Write and Edit Code, Build Reusable AI Workflows
  • Gemini Connected Apps (Extensions) [Verified] Re-checked support.google.com/gemini/answer/13695044 (still 200, live content). Confirms the connected-service list (Gmail, Calendar, Search, Flights, Hotels, Maps, YouTube, Spotify, YouTube Music, WhatsApp, Home/smart devices, Phone, Messages) and custom MCP-server apps; page states no per-plan pricing. Android's ability to use some apps (Phone, Messages, WhatsApp, device assistance) without Keep Activity on supports the "deepest integration" platform note. No claim changes Affects: Connect to External Systems
  • Gemini Past Chats (Memory) [Correction] Original URL (support.google.com/gemini/answer/14901455) now 404s; replaced with the live help article (answer/16598469, "Get personalization with memory of your past Gemini chats"), added as primary source. Google's settings path renamed Personal context → Personal Intelligence; talking point corrected. Added eligibility requirements confirmed on the article (18+, personal Google Account only, Keep Activity on; unavailable in Gems/Live chats) to Regional. Chrome platform corrected ❌ → ✅: the article states the feature works in "Gemini in Chrome (in countries where Gemini in Chrome is available)," matching this file's existing convention for that field (see Chat record) Affects: Remember Context Over Time
  • Gemini Project Astra [Verified] Re-checked deepmind.google/models/project-astra/. Still describes Astra as "a research prototype, being used and refined by a limited number of trusted testers," with screen sharing and video understanding already integrated into Gemini Live, and further Astra capabilities (Search, glasses) planned. Matches this record's deprecated status and talking point exactly. No claim changes
  • Grok Memory [Verified] Official primary source found and added — x.ai/grok (archive.org snapshot 2026-09-21; direct fetch blocked by Cloudflare) lists "Memory across chats — Remembers your preferences and past conversations" as a general, ungated feature, consistent with this record's free-tier-included claim. blog.memoryplugin.com retained as the only source for the Settings → Data Controls toggle detail, which x.ai/grok does not address Affects: Remember Context Over Time

2026-09-23

  • Microsoft Copilot Agent Builder [Correction] Restored Microsoft's February 2026 "What's New" post, removed 2026-04-23 after repeated link-checker flags. Those were false positives: the server answers HEAD with 400 while GET redirects to the live post (fixed in #640). It is the record's only source for the March 2026 claims: agents calling other agents as tools, and agents in Outlook, OneDrive and Teams communities. Restored at its current URL Affects: Build Reusable AI Workflows, Take Actions and Run Tools
  • Microsoft Copilot Copilot Connectors [Correction] Federated connectors checked against Microsoft Learn's federated connectors overview (updated 2026-09-23). The page no longer describes them as a preview: Microsoft-published federated connectors are enabled by default unless an admin disables them, and create/update/delete actions begin rolling out in early October 2026. Intercom is not in the current connector list and is removed from the talking point; Canva is added. Plan rows not re-verified, so Verified is unchanged Affects: Connect to External Systems
  • Microsoft Copilot Copilot Connectors [Correction] Restored Microsoft's February 2026 "What's New" post, removed 2026-04-23 after repeated link-checker flags. Those were false positives: the server answers HEAD with 400 while GET redirects to the live post (fixed in #640). It names the federated connectors in the talking point (plus Canva) and describes them as Public Preview across all tenants. Current preview/GA status was not re-verified Affects: Connect to External Systems
  • Gemini Connected Apps (Extensions) Record retitled Extensions → Connected Apps (Extensions) to match Google's current name for the feature. Implementation ID `gemini-extensions` is unchanged so API and page links stay stable Affects: Connect to External Systems
  • Gemini Connected Apps (Extensions) [Correction] Help article restored as URL and source. The 2026-03-14 "old URL 404" was a link-checker false positive: support.google.com answers HEAD with 404 while GET serves the page (fixed in #640). Its replacement, gemini.google.com/extensions, is a sign-in page with no readable content and is dropped as a source. Google now calls the feature Connected Apps, and the article adds custom apps connected through an MCP server. The article states no per-plan terms, so plan rows and Verified are unchanged Affects: Connect to External Systems
  • Gemini Gemini Live (Voice Mode) Camera and screen sharing recorded (capability `see-images-and-screens` added). Google's Live help article documents both on Android and iOS, with no plan restriction stated. These capabilities were previously credited to the Project Astra record, which is now folded into this one (#637) Affects: See Images and Screens, Hear Audio and Speech, Speak Back in Real Time
  • Gemini Project Astra Status ga → deprecated; folded into Gemini Live (#637). Google DeepMind's Project Astra page now calls it a research prototype (waitlist, trusted testers) and says its screen sharing and video understanding have been integrated into Gemini Live. The consumer camera/screen feature is recorded on the Gemini Live record, which now carries `see-images-and-screens`. This record's capabilities are cleared to avoid double-counting; implementation ID `gemini-project-astra` is kept

2026-09-11

  • Gemini Agentic App Actions (Android) [Verified] Plan rows renamed AI Pro → Google AI Pro to match the file's Pricing table (the primary plan vocabulary per scripts/lib/claims.js). No new rows: Google's subscriptions page makes no per-plan statement about this feature Affects: Connect to External Systems, Take Actions and Run Tools
  • Gemini AI Studio [Verified] Plan rows renamed AI Pro → Google AI Pro to match the file's Pricing table (the primary plan vocabulary per scripts/lib/claims.js). No new rows: Google's subscriptions page makes no per-plan statement about this feature Affects: Write and Edit Code, Build Reusable AI Workflows
  • Gemini Audio Overview [Verified] Added Google AI Plus and Google AI Ultra rows from Google's subscriptions page (gemini.google/subscriptions/); existing rows renamed to Google AI Pro
  • Gemini Canvas [Verified] Added Google AI Plus and Google AI Ultra rows from Google's subscriptions page (gemini.google/subscriptions/); Canvas is listed in the $0 plan card Affects: Make and Edit Documents, Write and Edit Code
  • Gemini Chat [Verified] Talking point plan names normalised to Google AI Pro / Ultra to match the availability rows; no factual change Affects: Write and Explain, Use It on My Surfaces
  • Gemini Deep Research [Verified] Added Google AI Plus and Google AI Ultra rows from Google's subscriptions page (gemini.google/subscriptions/); Deep Research is listed in the $0 plan card Affects: Do Multi-Step Research, Search the Web
  • Gemini Connected Apps (Extensions) [Verified] Plan rows renamed AI Pro → Google AI Pro to match the file's Pricing table (the primary plan vocabulary per scripts/lib/claims.js). No new rows: Google's subscriptions page makes no per-plan statement about this feature Affects: Connect to External Systems
  • Gemini Gemini Advanced [Verified] Status ga → deprecated: the name no longer appears on gemini.google/about, /subscriptions or the Live and desktop pages, and help article 14577972 returns 404. Free row ❌ → ⚠️ per Google's subscriptions page (gemini.google/subscriptions/) ("Varying access to 3.1 Pro"). Pro/Ultra notes and talking point prices corrected (Ultra from $99.99/mo; 5 TB / 20 TB+). No Plus row, per the 2026-04-07 decision
  • Gemini Gems (Custom Chatbots) [Verified] Added Google AI Plus and Google AI Ultra rows from Google's subscriptions page (gemini.google/subscriptions/); Gems is listed in the $0 plan card Affects: Build Reusable AI Workflows
  • Gemini Imagen [Verified] Added Google AI Plus and Google AI Ultra rows from Google's subscriptions page (gemini.google/subscriptions/); image generation and editing is listed in the $0 plan card Affects: Generate Images
  • Gemini Gemini Live (Voice Mode) [Verified] Gating corrected paid → free: Google's subscriptions page lists Gemini Live in the $0 plan. Availability expanded from two rows to the full Free / Google AI Plus / Pro / Ultra lineup with official usage multipliers (2x, 4x, 5x–20x) and Live-specific tier features from Google's 2026-08-26 post (Daily Brief on Plus+, Spark on Pro+). Web ❌ now cites Google help directly. Three official sources added. Resolves #632 questions 1 and 2 for this record Affects: See Images and Screens, Hear Audio and Speech, Speak Back in Real Time
  • Gemini Nano Banana Pro [Verified] Free row ❌ → ⚠️: Google's subscriptions page (gemini.google/subscriptions/) lists "limited access to Nano Banana Pro" via Google Flow in the $0 plan; Google AI Plus and Google AI Ultra rows added. Gating stays paid — full Gemini-app access still starts at Google AI Pro Affects: Generate Images
  • Gemini NotebookLM [Verified] Added Google AI Plus and Google AI Ultra rows from Google's subscriptions page (gemini.google/subscriptions/), which now calls the product "Gemini Notebook" (record title kept; implementations/index.yml binds to the heading) Affects: Organize Work in Projects, Use Files I Provide, Read Text and Documents, Do Multi-Step Research
  • Gemini Project Astra [Verified] Plan rows renamed AI Pro → Google AI Pro to match the file's Pricing table (the primary plan vocabulary per scripts/lib/claims.js). No new rows: Google's subscriptions page makes no per-plan statement about this feature
  • Gemini Veo (Video Generation) [Verified] Added Google AI Plus row — video generation is included at Plus per Google's subscriptions page (gemini.google/subscriptions/); Limits now carry Google's published Flow credit allocations; talking point prices corrected Affects: Generate Video
  • Gemini Gemini in Workspace (Docs, Sheets, etc.) [Verified] Added Google AI Plus row (Gmail and Vids, not Docs) from Google's subscriptions page (gemini.google/subscriptions/); talking point prices corrected — Ultra starts at $99.99/mo, not $249.99 Affects: Make and Edit Documents, Read Text and Documents, Connect to External Systems, Use Files I Provide, Use It on My Surfaces
  • Grok Grok Imagine (Video Generation) Per-tier daily generation counts (~3/day free, 50/100/500 per day paid) flagged as unsourced — their only backing was the unofficial supergrok.online source removed in #620, and no official xAI source publishes per-tier consumer video limits. docs.x.ai documents only API per-second pricing, 15s max duration and resolution caps; paid plans are reported to draw on a shared usage pool across Chat/Imagine/Voice/Build. Restructure deferred to the grok.md tier-taxonomy issue. Verified deliberately not bumped Affects: Generate Video
  • Grok Memory [Verified] Dead source compareclaw.com removed (HTTP 402, Vercel DEPLOYMENT_DISABLED — entire domain offline). No official xAI page describing consumer memory was found, so the record now rests on a single third-party source. Talking point corrected — a memory toggle exists under Settings → Data Controls (two independent reports); the earlier claim that memory cannot be inspected or edited was itself unsourced. Verified not bumped: no primary source Affects: Remember Context Over Time

2026-09-07

  • Microsoft Copilot Chat [Correction] The prior universal Copilot Pro retirement and 2026-08-01 support-end assertions were unsupported. Microsoft closed new sales, while existing subscriptions can continue. Removed the legacy unavailability row; feature entitlements for continuing subscribers require separate review. Checked/Verified dates are unchanged. Affects: Write and Explain, Use It on My Surfaces
  • Microsoft Copilot Image Generation (Designer) [Correction] The prior universal Copilot Pro retirement and 2026-08-01 support-end assertions were unsupported. Microsoft closed new sales, while existing subscriptions can continue. Removed the legacy unavailability row; feature entitlements for continuing subscribers require separate review. Checked/Verified dates are unchanged. Affects: Generate Images
  • Microsoft Copilot Model Access [Correction] The prior universal Copilot Pro retirement and 2026-08-01 support-end assertions were unsupported. Microsoft closed new sales, while existing subscriptions can continue. Removed the legacy unavailability row; feature entitlements for continuing subscribers require separate review. Checked/Verified dates are unchanged.
  • Microsoft Copilot Copilot in Office Apps [Correction] The prior universal Copilot Pro retirement and 2026-08-01 support-end assertions were unsupported. Microsoft closed new sales, while existing subscriptions can continue. Removed the legacy unavailability row; feature entitlements for continuing subscribers require separate review. Checked/Verified dates are unchanged. Affects: Make and Edit Documents, Read Text and Documents, Connect to External Systems, Use It on My Surfaces
  • Microsoft Copilot Copilot Vision [Correction] The prior universal Copilot Pro retirement and 2026-08-01 support-end assertions were unsupported. Microsoft closed new sales, while existing subscriptions can continue. Removed the legacy unavailability row; feature entitlements for continuing subscribers require separate review. Checked/Verified dates are unchanged. Affects: See Images and Screens
  • Microsoft Copilot Copilot Voice [Correction] The prior universal Copilot Pro retirement and 2026-08-01 support-end assertions were unsupported. Microsoft closed new sales, while existing subscriptions can continue. Removed the legacy unavailability row; feature entitlements for continuing subscribers require separate review. Checked/Verified dates are unchanged. Affects: Hear Audio and Speech, Speak Back in Real Time

2026-09-05

  • Microsoft Copilot Agent Builder [Verified] Copilot Pro row replaced by the Microsoft 365 consumer plans, all marked unavailable: agent authoring remains a commercial Copilot capability. Recorded that Microsoft 365 Premium ships prebuilt Analyst and Researcher agents, which is agent consumption rather than authoring. Affects: Build Reusable AI Workflows, Take Actions and Run Tools
  • Microsoft Copilot Chat [Verified] Copilot Pro row replaced by the Microsoft 365 consumer plans after Copilot Pro's retirement; free tier confirmed as no-sign-in basic chat with sign-in unlocking history, longer conversations, image creation and voice. Noted that paid plans buy usage rather than a different model, and that Premium adds the Analyst and Researcher agents. Refreshed the redirected support source. Affects: Write and Explain, Use It on My Surfaces
  • Microsoft Copilot Copilot Connectors [Verified] Copilot Pro row replaced by the Microsoft 365 consumer plans, all marked unavailable: connectors remain a commercial Copilot capability requiring a Microsoft 365 Copilot license. Affects: Connect to External Systems
  • Microsoft Copilot Image Generation (Designer) [Verified] Copilot Pro retired and replaced by Microsoft 365 Premium. Per-plan credits re-cut from Microsoft's AI credits support page: 60 credits/mo on Personal and Family, usage beyond the standard limit on Premium. Free-tier allowance corrected to "not published" — the previous 15 credits/mo figure described legacy Classic Microsoft 365, not free Copilot. Affects: Generate Images
  • Microsoft Copilot Model Access Renamed from "GPT-4 Access" to "Model Access". Microsoft no longer sells a named model tier, so the old name described a product that stopped existing; the record covers model access and the entitlements attached to it. Implementation id changed from copilot-gpt-4-access to copilot-model-access, which changes the record's permalink anchor.
  • Microsoft Copilot Model Access [Verified] The GPT-4 framing is obsolete: Microsoft's plan documentation no longer sells a named model tier, and publishes no free-versus-paid model split. Gating changed from paid to mixed, since the paid differentiator is usage limits and advanced features rather than model access. Copilot Pro rows replaced with the Microsoft 365 consumer plans. Dead Copilot Pro store source replaced with Microsoft support pages. Multi-model routing across OpenAI and Microsoft MAI models is secondary reporting, not first-party confirmed. Record is a candidate for renaming away from "GPT-4 Access".
  • Microsoft Copilot Copilot in Office Apps [Verified] Copilot Pro retired and replaced by Microsoft 365 Premium; Office app integration confirmed as included in Microsoft 365 Personal, Family and Premium, limited to the subscription owner and metered by AI credits. Replaced the dead Copilot Pro store source with Microsoft's free-vs-subscription and AI credits support pages. Affects: Make and Edit Documents, Read Text and Documents, Connect to External Systems, Use It on My Surfaces
  • Microsoft Copilot Copilot Vision [Verified] Gating reverted from free to paid: Microsoft now requires a Microsoft 365 Personal, Family, or Premium subscription for Copilot Vision. Status moved from preview to ga. Added per-plan daily limits (10 min Personal/Family, 15 min Premium) and commercial Entra ID access. Platforms expanded well beyond Edge: iOS and Android now supported, Windows shares any app, macOS is a native app. Regional widened from US-only to all supported Copilot regions and languages. Affects: See Images and Screens
  • Microsoft Copilot Copilot Voice [Verified] Copilot Pro retired (off sale October 2025, support ended 2026-08-01) and replaced by Microsoft 365 Premium; availability rows re-cut to Free / M365 Personal / Family / Premium. Free tier corrected from "Standard" limits to free-subject-to-capacity, sign-in required. Added Microsoft's published daily voice allowances (30 min/day Personal and Family, 60 min/day Premium). Added language and government-cloud caveats to regional note. Status remains ga; platform flags unchanged. Affects: Hear Audio and Speech, Speak Back in Real Time

2026-09-04

  • ChatGPT Sora (Video Generation) [Verified] platform: no longer available on Windows Affects: Generate Video

2026-08-31

2026-08-25

  • ChatGPT Codex (Code Agent) [Verified] Gating paid → free — OpenAI documents Codex as included on all ChatGPT plans including Free and Go; corrects contradiction with the availability table Affects: Write and Edit Code, Take Actions and Run Tools
  • Grok Grok Imagine (Video Generation) [Verified] Source swap — unofficial supergrok.online (CI link-check timeout) replaced with official xAI Imagine and video-generation docs Affects: Generate Video

2026-08-15

2026-07-25

  • ChatGPT Advanced Voice Mode [Verified] Free tier gained advanced voice (GPT-Live-1 mini); gating paid → mixed; macOS desktop voice returned Jul 23 2026 for paid plans; GPT-Live replaces Advanced Voice Mode as the default experience Affects: Hear Audio and Speech, Speak Back in Real Time

2026-07-22

  • ChatGPT Canvas [Verified] Status ga → deprecated. Canvas removed from GPT-5.5 Instant/Thinking (replaced by inline writing/code blocks); still default in GPT-4o for all users; paid legacy-model access sunsets 2026-08-26. Availability notes updated; talking point rewritten. Affects: Make and Edit Documents, Write and Edit Code

2026-06-18

  • Grok Real-time X/Twitter Data [Verified] Gating corrected paid→free (free tier has real-time access, ~10 msgs / 2h); SuperGrok Lite ($10/mo) and SuperGrok Heavy ($300/mo) added; terminal flipped to ✅ for Grok Build CLI (SuperGrok beta); Tesla platform added; URL normalized to grok.com Affects: Connect to External Systems

2026-04-08

  • ChatGPT DALL-E Image Generation [Verified] DALL-E 3 replaced by GPT Image 1.5 in ChatGPT (March 2025); DALL-E 3 API deprecated May 12, 2026; status changed to deprecated Affects: Generate Images
  • Claude Connectors [Verified] Gating changed from paid to free — Free plan now has full remote MCP connector access (not just desktop extensions + GitHub) Affects: Connect to External Systems
  • Claude Extended Thinking [Verified] Gating changed from paid to free — Free plan now has limited Extended Thinking access; Pro upgraded from limited to full access

2026-04-07

  • Gemini Deep Research [Verified] API platform row corrected to ❌ — Deep Research is a consumer UI feature with no dedicated Gemini API endpoint; two independent sources found no evidence of API access Affects: Do Multi-Step Research, Search the Web
  • Gemini Gemini Advanced [Verified] Removed AI Plus tier (two April 2026 sources confirm Gemini Advanced access is Pro/Ultra only; Plus not mentioned); added AI Ultra row; updated talking point
  • Gemini Imagen [Verified] Gating changed from paid to free; Imagen 4 now available to all Gemini users; free tier ~10-20/day, AI Pro ~100-150/day; model updated from Imagen 3 to Imagen 4 Affects: Generate Images

2026-03-25

2026-03-23

2026-03-21

  • ChatGPT Chat [Verified] GPT-5.4 Mini and Nano released: Mini available in ChatGPT (Free/Go via Thinking menu), API, and Codex; Nano is API-only; Mini $0.75/$4.50 per 1M tokens, Nano $0.20/$1.25 per 1M tokens; 400K context; vision support Affects: Write and Explain, Use It on My Surfaces

← Browse capabilities