What changed in AI tools: week 41, 2026
5–11 Oct 2026 · 32 updates from 22 tools · 1 verified price or limit change. Pricing and limits first.
Pricing 5
- The X Subscription merges Grok, Grok Bot, Cursor and X Premium into one plan
Plans run Free, Plus $8, Premium $30, Super $100 and Ultra $200, with a shared Grok usage pool; third-party models in Cursor bill at API list prices via top-ups.
Source: x.aiCurrent Grok plans & limits - Amp now runs on your Claude Pro or Max subscription
A new Claude Code mode uses the Claude Agent SDK and bills to a linked Claude Pro/Max plan, free for everyone, with orbs, runners and thread sharing still working.
Source: ampcode.comCurrent Amp plans & limits - Orgs can bill Copilot code reviews to a cost center instead of member quotas
Reviews still bill to the member's own Copilot entitlement by default; owners can switch them to the organization to avoid exhausting member quotas.
Source: github.blogCurrent GitHub Copilot plans & limits - HeyGen Video API adds 1080p and 2K output, both billed at 3× the 768p rate
768p stays the default; the multiplier also applies to reference-video seconds.
Source: docs.heygen.comCurrent HeyGen plans & limits - Perplexity Agent API presets switch to Fast Search, cutting cost to $1 per 1K
Low, medium and high presets now bill web_search at $1.00 instead of $2.50 per 1,000 calls; xhigh is unchanged.
Source: docs.perplexity.aiCurrent Perplexity plans & limits
Limits 2
- Lovable paid plans can set credit usage limits, alerts and blocks
Admins can cap credits per workspace, project or member, and on Business/Enterprise per group or token, with alerts or hard blocks.
Source: docs.lovable.devCurrent Lovable plans & limits - HeyGen API Enterprise rate limits raised up to 4×
Enterprise workspaces get 1,000 reads/min (was 300) and 30 writes/sec (was 10); Pay-As-You-Go and subscription limits are unchanged.
Source: docs.heygen.comCurrent HeyGen plans & limits
Launch 2
- Kilo Desktop launches with 500+ models and local model management
The app combines agentic coding and data science workflows with built-in management of local models and conda environments.
Source: blog.kilo.ai - ElevenAgents adds Reception for small businesses and Architect for agent tuning
Reception is an AI receptionist that answers calls and books jobs; Architect drafts improvements and tests them in simulated conversations.
Source: elevenlabs.io
Model 4
- JetBrains releases Mellum2.1, an open 12B coding-agent model
Still 2.5B active parameters under Apache 2.0, but reinforcement learning means it can now explore a repo, edit files and check its own changes.
Source: blog.jetbrains.com - Mistral Large 4 preview: 1T-parameter open-weight model, weights due end of month
The natively multimodal model has 52B active parameters and is in public preview on Mistral Studio.
Source: mistral.ai - GPT-6 with Intelligent UI rolls out to everyone in ChatGPT
OpenAI is rolling GPT-6 out globally in ChatGPT, with Intelligent UI that answers with visuals and interactive elements.
Source: openai.com - Claude Haiku 5.5 launches; Max and Team plans gain monthly API credits
Anthropic calls Haiku 5.5 its cheapest, fastest and most capable small model; Max and Team plans now include monthly API credits for the Claude Platform.
Source: support.claude.com
Feature 8
- Amp follows up on its own pull requests: comments, reviews and CI land in the thread
New first-class GitHub webhooks let an agent subscribe to a PR's comments, reviews and CI, or spawn a new orb for every new issue or PR.
Source: ampcode.com - ElevenLabs Brand Kit applies your logo, fonts and colors automatically
Add brand assets once and every new creation pulls them in.
Source: elevenlabs.io - Midjourney tests Thinking Mode and launches shared folders
On the alpha site, V8.2 jobs can be rerun with thinking to fix missed prompt details, and folders can be shared with collaborators or viewers.
Source: updates.midjourney.com - Free and Pro Lovable plans can publish apps invite-only
Publishing to a whole workspace or to groups still requires Business or Enterprise.
Source: docs.lovable.dev - Claude Docs, Slides and Design leave beta on every plan, including Free
Claude Dashboards and Claude Motion enter beta at the same time.
Source: claude.com - Warp Factories now works with Microsoft Teams and Azure DevOps
Warp Factories can now be used from Microsoft Teams and Azure DevOps.
Source: warp.dev - Raycast's Auto model gets Lightweight, Balanced and Flagship modes
Balanced, the default, uses about 70% fewer credits than Flagship, which needs Pro+ or Max.
Source: raycast.com - Cursor's iOS app can now watch and message agents running on your computer
Remote Control is on by default except for Enterprise orgs; agents keep running locally.
Source: cursor.com
Releases 6
- Kiro CLI V3 becomes the default, with background workflow agents
V3 brings the IDE's agent harness to the terminal and adds cloud sessions, spec mode and tangent mode.
Source: kiro.dev - Raycast 2.5–2.8: extension model providers, Send Element to AI, big speedups
Extensions can supply AI models, Send Element/Window to AI share screen context, and File Search is up to 6× faster.
Source: raycast.com - Mistral Vibe 2.26: onboarding wizard, teleport to web, voice mode
The Rust CLI gains first-run onboarding with keyring storage, /teleport to Vibe Code Web, /voice and /proxy-setup.
Source: github.com - Copilot week of 5 Oct: Haiku 5.5, local sandboxing GA, split accounts in the app
Haiku 5.5 reached Pro and above, local sandboxing went GA at no extra cost, and the Copilot app can use separate accounts for license and repos.
Source: github.blog - Claude Code 2.1.291–296: Haiku 5.5 default, blocking hook failures, sub-agent effort
2.1.293 makes Claude Haiku 5.5 the default Haiku ($0.10/$0.50 per Mtok, 1M context); hooks gain onFailure: "block" and the Agent tool gains an effort parameter.
Source: github.com - Replit Desktop for Windows enters private preview
Agent builds run in isolated local sandboxes (Microsoft MXC, NVIDIA OpenShell); chats can now read and task your other projects.
Source: docs.replit.com
Incident 2
- Augment major outage takes services down for over five hours
From 7:30 am to 1:06 pm PDT on 9 Oct multiple services were unavailable; Augment blamed permissions after a recent infrastructure migration.
Source: status.augmentcode.com - Outage: elevated errors across ChatGPT, Codex and the OpenAI API
OpenAI's status page listed most components as affected, including Conversations, Codex, Responses and the Agents API, before reporting full recovery.
Source: status.openai.com
Other 3
- Augment Code is selling Cosmos, Auggie CLI and its Context Engine to Harness
Augment agreed to sell these assets to Harness, and the team behind the products is moving to Harness to keep building them.
Source: augmentcode.com - v0 personal accounts become team workspaces
Chats, projects and settings moved automatically, and every account can now invite collaborators.
Source: v0.app - ChatGPT ads get a visual format and expand to Southeast Asia and Taiwan
OpenAI added a visual ad format and measurement tools in October, after expanding ChatGPT Ads to Southeast Asia and Taiwan (60+ countries) in September.
Source: openai.com
Verified price & limit changes
-
benefit addedClaude Max now includes monthly API credits ($100 on Max 5x, $200 on Max 20x)
Anthropic's platform release notes: 'Claude Max and Team plans now include monthly API credits.' Credits don't cover interactive Claude Code and don't roll over.