AI coding tools news
November 11, 2026 · 28 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.
Codex v0.167.2 adds incremental tool updates on Vertex AI
Codex version 0.167.2 now sends tool updates incrementally by default when using Google Cloud Vertex AI. Users can opt out by setting incrementaltoolslegacygcp = false under the [features] section in config.toml.
Kiro CLI 3.2.1 adds output detail settings and sign-in fixes
Kiro CLI 3.2.1 adds a tool call detail setting in /settings ranging from one line to full output. It fixes sign-in on Linux hosts without a graphical session and with Identity Center flags. The update also maintains /model fallback settings and resolves workflow, spec, Skills, file search, permissions, and session reload issues.
Google Gemini Decisions API launches with Gemini Decision 1 model
Google launched the Gemini Decisions API with the Gemini Decision 1 model. It costs 10 cents per million input tokens and output is free. The model supports text, images, audio, video, and PDFs. Jev remains cheaper at 4.2 cents per million input tokens.
AWS launches open‑source emulator for local testing of services
AWS released an official open‑source emulator that lets developers run a local clone of many AWS services without a paid API key. The emulator is written in Rust and was built by a swarm of AI agents, accumulating 29,684 commits in its first three weeks. It currently requires Docker to run on macOS or Windows.
Nvidia donates OpenShell runtime to CNCF
Nvidia donated its open-source OpenShell runtime for securing and governing AI agents to the Cloud Native Computing Foundation. The donation was announced this week.
Anthropic releases coding model that outperforms Opus 5.5
Anthropic released a coding model that beats Opus 5.5 on two major benchmarks and can complete certain difficult tasks.
Fractile raises $720M round, becoming top‑funded AI chip player
Fractile has raised a $720 million funding round. The company is now one of the top‑funded players in the race for purpose‑built AI chips.
Anthropic Claude Fable 5.2 builds shooter costing $550
Claude Fable 5.2 was given a single prompt to build a first‑person shooter. The process ran for eight hours and cost $550.
Claude Code update: preserved thinking and model setting changes
Preserved thinking changes how the Messages API handles thinking blocks to protect against distillation. The model, effort, and thinking settings have been changed.
AWS launches free open-source AWS Emulator built by AI agents
AWS launched an emulator for AWS called AWS Emulator. It is free and open source. The emulator was built by AI agents and AWS now maintains it.
Antigravity v1.5.5 adds font size setting, split view, download approval, swarms preview and Gemini 4 Carbon support
This release adds a font size setting, split‑view plugin windows, and requires approval before the browser agent downloads a file. It also brings call summaries to voice mode, Google Cloud project autocomplete to Pay‑As‑You‑Go setup, and moves per‑project configuration to <project>/.gemini/config.json. Version v1.5.5 introduces a preview of agent swarms with the /swarm command, adds Gemini 4 Carbon model support, enforces HTTPS for remote enterprise model gateways, and includes plugin import fixes and other improvements.
MCP file‑upload issue fix in development
The Model Context Protocol (MCP) has had a file‑upload problem for two years. A fix is now taking shape.
Adobe Firefly launches all‑in‑one AI video editor
Adobe introduced Firefly video editor, an AI‑powered all‑in‑one creative tool for transforming videos. The article provides no pricing or usage limit details.
OpenAI quarterly revenue exceeds $10 B for first time
OpenAI reported quarterly revenue of $10 billion, an almost 50% jump from $6.7 billion in the prior quarter.
Google launches Tabletop AI tool for custom collaboration hubs
Google introduced an experimental AI tool called Tabletop that creates customized collaborative hubs. Users can describe desired interactive tools and the platform generates them. The interface resembles workplace apps such as Asana and Monday.
Introducing the Extractions API
A new API for pulling typed fields out of documents and images was announced. It returns field names, values, spans and confidence scores. Three models are available at launch. On Cloudflare's Clairic the same request can read text and images and return numbers, dates, lists, estimates and forecasts.
Anthropic releases IPO roadshow video detailing multi‑chip strategy
Anthropic posted a video from its IPO roadshow. The video explains that Claude is trained and served across AWS Trainium, Google TPUs, NVIDIA GPUs and AMD GPUs as part of a multi‑chip strategy. No additional pricing or quota details are provided.
Plurai aims to improve AI agents using self‑driving car approach
Plurai wants to help AI agents work better using the same approach that helped get self‑driving cars to roads.
Google DeepMind unveils AlphaGenome 2 model
Researchers at Google DeepMind unveiled AlphaGenome 2, an AI model that studies how DNA changes affect cells. The model provides a more detailed view of cellular impacts and may aid disease research and treatment development.
Databricks launches revised Genie AI agent
Databricks introduced a revised Genie AI agent. The CEO Ali Ghodsi uses the agent to gather unfiltered feedback from employees. The announcement was made in a TechCrunch article on November 11, 2026.
Baidu Wenxin new model to be open‑sourced
The new Wenxin model is announced to be open‑sourced soon.
Multiple AI Model Vendors Offer 50% Discount on Tmall
Several large‑model providers – Zhipu, MiniMax, Qianwen Office, Alibaba Cloud and Qoder – are offering a 50 % discount on Tmall. The promotion is announced for November 11.
Cogent and OpenAI build cyber defense factories for enterprises
Cogent combines OpenAI models with enterprise context to help automatically remediate vulnerabilities with agents.
ChatGPT Plus offers free year to veterans and service members
More than 20,000 veterans and transitioning service members have claimed a free year of ChatGPT Plus. The offer provides them with full access to the Plus tier for one year at no cost.
Kiro CLI 3.2 adds full-screen Mermaid viewer and broader MCP sign‑in
This release adds a full‑screen viewer for Mermaid diagrams that exceed terminal width and enables sign‑in to additional MCP servers without OAuth configuration. It also allows one extra iteration or finishing a workflow loop that has used all its iterations, notifies when workflows are unavailable in autonomous mode, and prompts before a V2 shell command modifies XDGCONFIGHOME or HOME. Several fixes were made for file search, custom agents, workflows, specs, hooks, clipboard, help text, and session handling.
Kiro Enterprise Management preview adds unified policy for new US East profiles
Kiro Enterprise Management is in preview for newly created Kiro profiles in US East (N. Virginia). Administrators write one policy in the AWS console and it is applied to the Kiro IDE, CLI, Web, and Crew LTS for all developers on the profile. Kiro IDE requires version 1.4.154 or later and Kiro CLI requires version 3.1.0 or later. Support for existing profiles and European regions will be added later.
OpenRouter blog explains MCP vs CLI vs Skills cost
The blog post explains that each tool an agent uses sends its definition to the model on every request, incurring input token costs. A Model Context Protocol (MCP) server can add tens of thousands of tokens to the context window. The alternatives are a command‑line tool (CLI) or a skill.
GitHub App support for Enterprise Copilot seat management
Enterprise GitHub Apps can now call the enterprise‑level Copilot seat management APIs to control which users and enterprise teams are assigned Copilot seats. The seat management APIs are accessible to these apps.