AI coding tools news
November 9, 2026 · 26 stories
The day's news for developers who use AI coding tools, picked and summarised from official changelogs, the media and Hacker News. Resets and limit changes are on the reset tracker.
JPMorganChase expands Claude Code deployment to more employees
JPMorganChase has expanded the rollout of Claude Code to additional employees across the firm.
Google AI Studio Adds Maps Platform Integration for App Building
Google AI Studio now enables building map-based applications directly from prompts using the Maps Platform. The integration covers places, routes, 3D views, and environmental data.
Meta Launches Muse Ember Model for Agent Tasks
Meta introduces the Muse Ember model featuring multimodal reasoning and a 1M-token context window for repetitive coding and agent tasks.
Anthropic launches Claude Fable 5.2 and Mythos 5.2 models
Anthropic releases Claude Fable 5.2 with faster output and higher technical benchmark scores. Mythos 5.2 is restricted to verified groups.
OpenAI Introduces Control Room for AI Workload Monitoring
OpenAI introduces Control Room, a central system designed to track model workloads across research infrastructure and rapidly stop them when safety or security concerns arise.
OpenAI Introduces Control Room for AI Workload Monitoring
OpenAI launches Control Room to track model workloads across its research infrastructure and rapidly stop them during safety or security concerns.
GitHub Copilot CLI 1.0.109-0 adds confirmation prompts and permission controls
Improved – asks for confirmation before /share or /export follows symlinks when saving local files. URL permission prompts let you approve every subdomain for the current session. Read‑only shell scripts ask for permission less often, while scripts that change git refs or read files through hidden paths always get reviewed. Fixed – updated description texts for beepOnSchedule, companyAnnouncements, and help config messages.
Antigravity CLI 1.5.2 adds cloud plugin skills, macOS Option shortcuts, security hardening
Antigravity CLI 1.5.2 adds cloud plugin skills to remote‑agent sessions, making them appear in the / menu and usable via /skills. It enables native macOS Option‑key shortcuts (e.g., alt+m) without extra configuration. Security hardening includes tighter permission boundaries, changes to --dangerously-skip-permissions, and stricter handling of protected commands and plugin agents.
Claude Code v2.1.321 adds Claude Fable 5.2 as default model
Added Claude Fable 5.2 (claude-fable-5-2) as the default Fable model with 1 M context, priced at $8 per Mtok for standard usage and $40 per Mtok for enterprise, plus $0.20 per Mtok for cache reads. Added a --settings <file> option to claude remote‑control to apply a settings file to every session. Fixed several security and sandbox permission bugs and corrected the usage‑credits note on Enterprise seats when switching to Fable.
ChatGPT Enterprise now available in AWS Marketplace
ChatGPT Enterprise is now available through AWS Marketplace, helping organizations simplify procurement via existing AWS purchasing and billing relationships.
Anthropic launches Claude Fable 5.2 and Claude Mythos 5.2 models
Anthropic announced two new models, Claude Fable 5.2 and Claude Mythos 5.2. The company says the models improve coding, reasoning, and vision capabilities. They also generate outputs faster than the previous 5.1 versions.
Apple adds $49.99 one‑time “Buy More Usage” for Logic Pro AI features
Apple updated Logic Pro with a “Buy More Usage” option for its AI assistant. The option is a one‑time purchase priced at $49.99. The change was noted after a support article outlined usage limits.
Google TPU Sync cuts inference latency by over 90%
Google announced its eighth‑generation TPU includes TPU Sync, an open‑source system that reduces inference latency by more than 90%. The announcement also introduced new TPU chips.
Instagram launches AI Effect Builder for custom Story filters
Instagram launched the Effect Builder that creates custom Story filters from AI prompts. It is available now for US residents who are over 18. The tool includes a prompt box for describing the filter idea and an option to attach a style reference.
Meta releases Muse Ember lightweight model for agent execution
Meta announced the Muse Ember model for agent execution. The model costs 10 cents per million input tokens, which is lower than competitors such as Haiku and Luna.
Baidu Wenxin reveals RSI progress and ERNIE Evolve use in upcoming Wenxin 5.5 Flash
On November 9, Baidu Wenxin disclosed its latest RSI progress. It stated that the ERNIE Evolve model has been used in developing its newest model. The upcoming Wenxin 5.5 Flash may be released soon.
Samsung paywalls AI features with $4.99 monthly plan
Samsung is now paywalling its best AI features. Leaked pricing points to $4.99 monthly with heavier features reserved for a costly Max tier.
ElevenLabs launches ElevenAgents Campaign Manager
ElevenLabs adds outbound campaign management to ElevenAgents, including follow-ups, callbacks, analytics and controls across phone and WhatsApp.
GitHub Copilot CLI 1.0.108 adds sandbox file enumeration and new /index command
The 1.0.108 release adds sandbox.userPolicy.filesystem.enumeratePaths in settings.json for Windows enumerate‑only directory grants. It also adds /index commands and enables managed shared tgrep indexing by default for eligible repositories. PDF attachments on Grok 4.7 and 4.6 now fall back to file references instead of failing. With the sandbox on, a local MCP server restarts after /cd to follow the new directory’s denied paths. File attachments that point to a named pipe, socket or device are now tagged as errors. Canvas tools now send instance IDs to opencanvas instead of free‑form text. The /model command now lists models when a 421 Misdirected Request is returned, re‑resolving the account endpoint as chat turns already do.
Disney and Adobe partner to build generative AI tools
Disney announced a multi‑year partnership with Adobe to develop proprietary artist‑led generative AI models for Disney’s production pipeline. Adobe will give Disney early access to its latest generative AI technologies. The collaboration will cover Disney’s theme parks, cruise line and ESPN assets.
Google adds new TPU 8tx and TPU 8i Max for AI workloads
Google expands its TPU lineup with the TPU 8tx and TPU 8i Max. The new chips support confidential computing and dual‑chip systems. They are aimed at agents handling large codebases and million‑token contexts.
Meta rolls out specialized models for Muse and Meta AI
Meta introduced Active Memory and Active Focus to Muse and Meta AI, reporting gains in context recall and task completion with specialized models.
Tencent HunYuan joins top open-source models – Hy4 tested with WorkBuddy multimodal agent
Tencent HunYuan was added to the list of leading open‑source models. The Hy4 model was evaluated together with the WorkBuddy multimodal agent, showing strong delivery capability.
Qianwen Office 2.0 released with Team Agent supporting multi‑person and agent collaboration
On November 9, Qianwen Office launched version 2.0. The update adds a Team Agent, the first feature that enables collaboration between multiple users and agents.
TokenRhythm releases NeoHorse-1.5-Flash Preview agent model
On November 5, TokenRhythm announced the release of its native agent model NeoHorse‑1.5‑Flash Preview.
Goldman Sachs pulls Claude Code from bankers and traders
Goldman Sachs is pulling Claude Code from its bankers and traders.