129 public catches this week · all weeks · the live wire

08-30URGENTct_logs

Two new subdomains, 'a.anthropic.com' and 'staging.identity.anthropic.com', appeared in Anthropic's certificate transparency logs.

08-30RTNhacker news

Claude is now appending session URLs to commit messages and PR descriptions by default, according to a fast-climbing Hacker News story.

08-28URGENTbinary

Anthropic's native binary for Claude Code on darwin-arm64 includes references to a new model `claude-3-5-haikuV`, hints at 'Agent Skills' and a 'Files API' moving out of beta, and new internal endpoints related to session management and connectors.

08-28FLASHgithub_releases+npm+pages

Claude Code (v2.1.251) introduces new hook events for model switching, live streaming of subagent tool calls, spend limits, prompt cache metrics, and CLI improvements, alongside numerous bug fixes.

08-28URGENTweb app

Anthropic's Claude platform console deployed new frontend code, revealing multiple new API endpoints and features related to organizational management, user accounts, service accounts, and workspaces, many with a 'beta' flag.

08-28URGENThacker news

A US judge ruled the Pentagon's blacklisting of Anthropic was unlawful.

08-28RTNgithub_org

Anthropic released a new public repository, 'claude-tag-wif-gateway-sample', providing a sample customer-side gateway for Claude Tag identity federation.

08-28RTNhacker news

A judge ruled the Trump administration's blacklisting of Anthropic was illegal, according to an HN story gaining traction.

08-28FLASHmodel_catalog

160 new model IDs added to the LiteLLM catalog, including potential new models from OpenAI (gpt-5.6-cyber), Anthropic (claude-fable-5, claude-opus-5, claude-sonnet-5), Google (gemini-3.1-flash-lite, gemini-3.1-pro, gemini-3.5-flash, gemini-3.7-flash, gemma-4-26B), and many others across various providers. ✓ beat the announcement by 7.2h

08-28URGENTbinary

Possible mention of a new Anthropic model, `claude-3-5-haiku`, found in a binary for `@anthropic-ai/claude-code-darwin-arm64` along with numerous internal error messages and feature flags.

08-28URGENTdocs

Anthropic updated their API changelog to announce new personal and service account API keys, and changes to beta API headers for files and skills in their SDKs.

08-28RTNweb app

Anthropic's Claude platform console deployed new frontend code, adding a new feature flag and endpoint related to an 'internal_tier' for organizations.

08-27URGENTgithub_org

Anthropic made a new public GitHub repository named 'formal-math'.

08-27RTNgithub

Anthropic's Python SDK v1.2.0 features GA shapes for beta files/skills namespaces and bug fixes for AWS Bedrock binary file uploads and tool handling.

08-27FLASHpypi

Anthropic's 1.2.0 Python SDK introduces new model IDs, "claude-opus-4-8" and "claude-haiku-4-5", along with a beta messages stream and refusal fallback middleware.

08-27URGENTnpm

Anthropic's SDK update includes new tools for file editing and range reading, improved webhook verification, and context window continuation summaries.

08-27FLASHgithub_releases+npm+pages

Anthropic's Claude Code v2.1.248 introduces a `--restricted` mode for enhanced security, per-agent prompt caching, self-hosted runner labeling, and various diagnostics and fixes.

08-27RTNweb app

Anthropic's Claude platform console deployed new frontend code including a new internal_tier feature flag and endpoint.

08-27RTNweb app

Anthropic's Claude platform console shows a new internal_tier endpoint and feature flag, possibly indicating new organizational tiering for platform access.

08-27FLASHbinary

Anthropic's latest binary includes references to an unannounced 'Claude Opus 5' model, a new plumbing tool, and expanded internal tool protocols. ✓ beat the announcement by 19.7h

08-27FLASHweb app

Anthropic's Claude.com frontend bundle includes new model_id strings: claude-3, claude-haiku-4, claude-opus-4, and claude-sonnet-4.

08-26FLASHgithub_releases+pages

Anthropic's Claude Code v2.1.247 release includes a new `SendFeedback` tool, `spinnerTipsOverride` for custom tips, a new `/claude-api cost-optimize` command, and updated Admin API coverage within the `/claude-api` skill, along with numerous bug fixes.

08-26URGENTsitemap

Anthropic quietly added 72 new pages to its docs sitemap, revealing an upcoming 'organization' API with features for managing API keys, external keys, federation, invites, rate limits, and service accounts.

08-26URGENTdocs

Anthropic's Admin API is now available in ant CLI and SDKs for several languages, including Python, TypeScript, and Java, covering organization management features.

08-26RTNgithub

Anthropic's Python SDK v1.1.0 adds a beta 'updates' thinking display mode, Organization API endpoints, and missing beta values.

08-26FLASHpypi

Anthropic 1.1.0 hints at two new model IDs, "claude-opus-4-8" and "claude-haiku-4-5", and introduces a "beta" feature flag across numerous organization API endpoints, suggesting new beta features for org management.

08-26FLASHnpm

Anthropic's SDK update 0.121.0 reveals extensive new organization and workspace management features, including API keys, external keys, federation, invites, service accounts, and user/member management, all under a beta flag.

08-26URGENTdocs

Anthropic's Compliance API for Claude is out of beta for Cowork and Claude Code sessions, and now includes Claude Science and Microsoft 365 session transcripts in beta for Enterprise organizations.

08-26RTNweb app

Anthropic's Claude platform added a new Stripe API endpoint related to payment intents, possibly for new billing or subscription features.

08-26FLASHmodel_catalog

LiteLLM's model catalog added 39 new model IDs, including several Databricks Claude variants, new RunwayML models like Veo and Hailuo, and numerous Together AI models from providers like Qwen, DeepSeek, Google, Meta, and NVIDIA.

08-26URGENTbinary

New internal model IDs and potential future Claude versions like 'opus-4-8', 'claude-3-5-haikuV', and 'claude-in-chrome' have appeared in the latest Anthropic binary. ✓ beat the announcement by 18.4h

08-25RTNgithub_releases+pages

Claude Code released version 2.1.246 with several bug fixes and minor feature additions, including auto mode classifier rules and improved transcript rendering.

08-25RTNhacker news

Anthropic has advised its San Francisco staff to work from home due to a potential security team strike, as indicated by a rapidly climbing Hacker News story.

08-25FLASHsitemap

Anthropic added new migration guides to their docs sitemap for unannounced models: Fable 5, Haiku 4.5, Opus 5, and Sonnet 5. ✓ beat the announcement by 0.9h

08-25URGENTbinary

Anthropic's Claude Code binary for macOS (ARM64) shows several new model IDs, beta feature flags, and API endpoints, suggesting upcoming new models and expanded CLI/tooling capabilities.

08-25FLASHgithub_releases+pages

Anthropic's Claude Code v2.1.245 fixes a startup crash on Linux distributions using glibc 2.44 and adds several new features for usage tracking, model management, and authentication.

08-25FLASHbinary

Anthropic's Claude Code binary shows new model IDs `claude-opus-4-7` and `claude-opus-4-8`, suggesting upcoming model releases or internal testing. ✓ beat the announcement by 42.4h

08-25URGENTgithub_releases+pages

Anthropic's Claude Code v2.1.243 release includes new features for loop usage breakdown, custom model selection, prompt caching, organizational pricing, and keyless sign-in, along with numerous bug fixes.

08-24FLASHsitemap

Anthropic's sitemap shows new docs for Claude "Fable 5" and "Mythos 5" models, plus numerous Opus and Sonnet 4.5/5/6/7/8 variations, hinting at a major model release or update, as well as a new 'browser-use-tool' capability. ✓ beat the announcement by 19.1h

08-24RTNweb app

Anthropic's Claude platform console deployed new frontend code, adding new endpoints related to user access, OAuth authorization redirects, and organization workspaces, plus an internal error feature flag.

08-30RTNgithub

OpenAI's Codex project has released an alpha pre-release for its Rust-based version 0.152.0.

08-30RTNweb app

ChatGPT web app update includes new strings referencing image features, plugins, and various OpenAI policy and help pages.

08-30FLASHweb app

ChatGPT's web app bundle update reveals several new model IDs, including mentions of 'gpt-5', 'gpt-5-pro', 'gpt-5-thinking', 'gpt-5.6-luna-wm', and various 'gpt-4o' variations, alongside numerous internal feature flags and new endpoints.

08-30RTNweb app

OpenAI is testing a feature flag for generating 'trending suggestions' within ChatGPT conversations.

08-29RTNgithub

OpenAI Codex released a new pre-release version for its Rust integration, v0.151.0-alpha.7.2.

08-29RTNgithub

OpenAI's Codex project published a new pre-release: rust-v0.152.0-alpha.1.

08-29RTNgithub

OpenAI's Codex Rust library (v0.151.0) adds configurable grace periods for tool discovery, allows extensions to process MCP tool results, and improves sandbox enforcement and plugin catalog handling.

08-29URGENThacker news

OpenAI published a statement regarding Cursor after its acquisition by SpaceX, which is gaining rapid attention on Hacker News.

08-29RTNgithub

OpenAI's Codex project published a new pre-release version: rust-v0.151.0-alpha.12.

08-29RTNweb app

A new feature flag named "generate_trending_suggestions" was added to the ChatGPT web app, possibly indicating an upcoming feature related to generating trending conversation suggestions.

08-28RTNgithub

OpenAI's Python library (v3.6.0) now includes 'compute_units' in API responses and Chat Completions usage.

08-28FLASHpypi

OpenAI's 3.6.0 Python package includes a new `gpt-4o-2024-08-06` model ID, beta real-time and responses connection features, and mTLS endpoints, suggesting new model and API capabilities.

08-28URGENTgithub

OpenAI's Codex project published a new pre-release version: rust-v0.151.0-alpha.11.

08-28URGENTgithub

OpenAI's Codex project published a new pre-release for its Rust-based version, 0.151.0-alpha.10.

08-28RTNgithub

OpenAI's Codex project has released a new pre-release version for its Rust component, `rust-v0.151.0-alpha.7.1`.

08-28FLASHmodel api

OpenRouter has added several new model IDs to its API, including new DeepSeek V4 versions, Google's Gemma 4 31B, Moonshot AI's Kimi K3, two OpenAI GPT-OSS models, Qwen 3.5 9B, and Z-AI's GLM 5.3 Flash.

08-28RTNgithub

OpenAI's Codex project released a new pre-release version: rust-v0.151.0-alpha.9, suggesting ongoing development in the Rust implementation. ✓ beat the announcement by 0.0h

08-28URGENTgithub

OpenAI's Codex project released a new pre-release version: rust-v0.151.0-alpha.8. ✓ beat the announcement by 4.9h

08-27RTNgithub

OpenAI's Codex project has published a new pre-release for its Rust-based version, 0.151.0-alpha.7, indicating ongoing development.

08-27URGENTnpm

OpenAI's latest SDK introduces file search chunking controls, a new `ranker` option, and X.509 credential validation messages, suggesting updates to RAG and security features.

08-27RTNgithub

openai/codex published a new pre-release: rust-v0.151.0-alpha.6, indicating ongoing development in their Rust-based tooling or SDK.

08-27RTNgithub

openai/codex published a pre-release: rust-v0.150.0-alpha.12.2, hinting at continued development.

08-27RTNgithub

OpenAI's Codex project published a new pre-release version: rust-v0.151.0-alpha.5.

08-27URGENTgithub

OpenAI's Codex project has released a new alpha pre-release for its Rust-based version, `rust-v0.151.0-alpha.4`. ✓ beat the announcement by 12.3h

08-27URGENTgithub

OpenAI's Codex repository published a new pre-release, rust-v0.150.0-alpha.12.1, indicating continued development on a Rust-based version of Codex. ✓ beat the announcement by 14.3h

08-27RTNgithub

A new release for openai/codex, rust-v0.150.1, includes a bug fix for remote compaction's token budget calculation by counting retained images.

08-27RTNgithub

OpenAI's Python library v3.5.0 makes function call output IDs optional, possibly simplifying certain API interactions.

08-27FLASHpypi

OpenAI's 3.5.0 Python package includes a reference to a new 'gpt-4o-2024-08-06' model ID and new 'realtime' and 'responses' beta API connections.

08-27RTNnpm

OpenAI's latest SDK hints at new 'ranker' options for file search, 'chunk_overlap_tokens', and 'static' chunking for file search tools, along with explicit 'code_interpreter' and 'file_search' tool types.

08-26RTNgithub

OpenAI's Codex project released a new pre-release version: rust-v0.151.0-alpha.3. This is an alpha release, indicating early development or testing.

08-26RTNgithub

OpenAI's Codex project published a new pre-release version: rust-v0.151.0-alpha.2.

08-26RTNgithub

OpenAI Python library v3.4.0 adds obfuscation to ChatCompletionChunk and project residency configuration with cost quantity units, alongside numerous bug fixes and documentation clarifications.

08-26FLASHpypi

OpenAI's latest SDK includes a new 'gpt-4o-2024-08-06' model ID and beta features for 'realtime' and 'responses' connections, hinting at potential streaming or interactive capabilities.

08-26URGENTnpm

OpenAI's v7.6.0 npm release adds new API endpoints, file search tool overrides with chunking controls, and a 'ranker' option, alongside numerous streaming-related error messages and local/regional API endpoints.

08-26RTNhacker news

An OpenAI blog post titled 'The Hugging Face incident and the road ahead' is gaining rapid traction on Hacker News.

08-26RTNgithub

OpenAI Codex `rust-v0.150.0` release introduces `@` mentions for tasks, enhanced `/copy` and `/rename` for terminal tasks, clickable Markdown links, and `Interrupt` hooks.

08-26RTNgithub

OpenAI's Codex project has released a new pre-release version: rust-v0.150.0-alpha.13. This appears to be an incremental alpha update for a Rust-based component of Codex.

08-26RTNgithub

OpenAI's Codex project has published a new pre-release version: rust-v0.150.0-alpha.12.

08-26RTNweb app

ChatGPT web app deployed a new experimental feature flag for generating trending conversation suggestions.

08-25RTNweb app

ChatGPT web app deployed a new feature flag related to generating 'trending suggestions' for conversations.

08-25RTNgithub

OpenAI's Codex project published a new pre-release for its Rust integration, version `rust-v0.150.0-alpha.11`.

08-25RTNgithub

OpenAI's Codex project released a new pre-release version: rust-v0.150.0-alpha.10, indicating continued development.

08-25RTNweb app

ChatGPT web app updated with new strings referencing staging, images, plugins, help, and pricing endpoints.

08-25FLASHweb app

OpenAI's ChatGPT web app has new strings indicating internal model development, including references to 'gpt-5' and 'gpt-4o-mini', and several 'gdrive' models, suggesting new capabilities or integrations. ✓ beat the announcement by 0.0h

08-25RTNgithub

OpenAI's Codex project released a new pre-release version: rust-v0.150.0-alpha.9, suggesting ongoing development in the Rust implementation of Codex.

08-25RTNweb app

ChatGPT web app adds a feature flag for experimental trending suggestions in conversations.

08-25RTNweb app

New frontend strings in ChatGPT web app suggest an experimental 'generate trending suggestions' feature flag and a staging endpoint.

08-24RTNgithub

OpenAI's Codex project has published a new pre-release, `rust-v0.150.0-alpha.8`, indicating ongoing development in its Rust implementation.

08-24RTNweb app

New strings related to images, plugins, OpenAI help, and ChatGPT pricing have been detected in the ChatGPT web app bundle.

08-24URGENThacker news

OpenAI's GPT 5.6 Sol model is showing a price reduction, according to an HN post, likely until at least November 21st. ✓ beat the announcement by 0.0h

08-24FLASHweb app

OpenAI's ChatGPT web app was updated, adding references to new models like "GPT-5", "GPT-5.6-preview", and various "GPT-4o-gdrive" versions, along with internal feature flags and numerous external service endpoints. ✓ beat the announcement by 18.2h

08-24FLASHlmarena

LMArena added two new models to its leaderboard: 'gpt-5.6-sol-search-xhigh' and 'grok-4.5-search'. ✓ beat the announcement by 18.2h

08-24RTNweb app

New strings in ChatGPT web app bundle indicate potential updates to image features, plugins, help resources, and pricing information.

08-24RTNweb app

OpenAI's ChatGPT web app updated its frontend code, adding new endpoints for images, plugins, and links to help and pricing pages.

08-24RTNweb app

ChatGPT web app update includes new feature flags, possibly related to dynamic content, internal rendering, and application tracking.

08-24FLASHweb app

ChatGPT's web app bundle includes mentions of "GPT-5", "gpt-5-pro", and various "gpt-4o-gdrive" models, along with new internal features and endpoints. ✓ beat the announcement by 33.1h

08-24FLASHweb app

OpenAI's ChatGPT web app was updated with new strings referencing 'GPT-5' and several 'gpt-4o' variations, alongside numerous internal model and feature flags. ✓ beat the announcement by 34.8h

08-24URGENTgithub

OpenAI Codex published a new Rust pre-release, `rust-v0.149.0-alpha.4.3`. ✓ beat the announcement by 38.0h

08-30FLASHmodel_catalog

43 new model IDs detected across various providers including new Gemini, Mistral, xAI Grok, and Databricks models.

08-29RTNgithub

Gemini CLI published a new pre-release version v0.59.0-nightly.20260829.g0bd1d4397, including a fix to enforce fail-closed workspace trust and filter mcpServers in restricted mode.

08-28FLASHhuggingface

Google pushed a new HuggingFace repository: google/timesfm-3.0-pytorch, which likely indicates a new model or release.

08-27FLASHmodel api

A new model ID, gemini-omni-1.1-flash, has appeared in Google Gemini's API, signaling a potential upcoming release.

08-27RTNgithub

Google Gemini CLI published a pre-release v0.59.0-nightly.20260827.g3c311beac with a fix for SSRF in MCP OAuth metadata discovery and authentication.

08-26URGENTgithub_org

Google Gemini made a new public repo for 'Jot Gemini Transcribe macOS', a smart dictation demo.

08-26FLASHmodel api

Two new model IDs, gemini-3.5-transcribe and gemini-3.5-transcribe-live, appeared in Google Gemini's API, indicating potential new transcription capabilities.

08-26RTNgithub

Google Gemini CLI published a new pre-release version v0.59.0-nightly.20260826.g64b5b79a6, indicating potential upcoming changes.

08-25RTNgithub

Gemini CLI published a new pre-release (v0.58.0-preview.0) with fixes for symlink evaluation, Docker isolation on macOS, and cancellation errors, alongside a stable release (v0.57.0) with new evaluation features and various core bug fixes.

08-25RTNgithub

Gemini CLI published a new pre-release `v0.56.0-nightly.20260825.g812f7a2bc` with fixes for cancellation errors, safety checker configuration, and history rollback optimizations.

08-24RTNgithub

Google Gemini CLI published a new pre-release version v0.56.0-nightly.20260824.g5411f113c.

08-27RTNdocs

Cursor's changelog added 'Grok Bot' to its Cloud and Mobile sections, possibly indicating a new integration or feature.

08-29FLASHhuggingface

Nvidia pushed a new HuggingFace repository: nvidia/DeepSeek-V4-Pro-0813-NVFP4, which appears to be a new model.

08-28FLASHmodel api

OpenRouter's API added several new Mistral AI model IDs, including "codestral-2508:batch," "ministral-8b-2512:batch," and updated versions of Mistral Large, Medium, and Small. These new batch inference models appear to be for scheduled or deferred processing.

08-28FLASHmodel api

OpenRouter's API now lists a new model ID: qwen/qwen3.8-2.4t-a95b:batch, suggesting a potential unannounced model.

08-27FLASHhuggingface

NVIDIA created two new HuggingFace repositories: one for a Qwen-related model variant and another called 'stormscope-meteosat'.

08-26FLASHmodel api

OpenRouter's API added a new model ID: qwen/qwen3.8-flash, suggesting an upcoming release.

08-26FLASHhuggingface

Qwen has pushed new HuggingFace repositories for what appears to be a new model, Qwen3.8-Flash-Next, with an FP8 version also available.

08-25FLASHlmarena

LMArena has a new model, qwen3.8-27b-code-image-to-webdev, on their leaderboard, potentially indicating a future release.

08-27FLASHdocs

Cursor Cloud Agents now allow starting projects from scratch without a SCM provider, offering in-browser live previews and direct Vercel publishing.

zai-org created new HuggingFace repositories for 'GLM-5.3' and 'GLM-5.3-BF16', likely indicating new model weights.

LMArena added a new anonymous model, "phoenix-qcs2," to its leaderboard, potentially indicating a stealth model release.

Nvidia has pushed a new HuggingFace repository named 'nvidia/Nemotron-3-Diarization-preview', potentially indicating an upcoming release of a new Nemotron-3 model variant focused on diarization.

LMArena added a new anonymous model, polaris-dqm3, to its leaderboard, likely an unreleased stealth model.

LMArena has added three new models to its leaderboard, including an anonymous model, "omni-1-1-eap-925h", and two named models, "glm-5.3-flash-text" and "glm-5.3-flash-vision". The anonymous model is likely a stealth test.

Microsoft created a new HuggingFace repository called 'SQuadGen'.

LMArena has added a new anonymous model, farris-ees5, to its leaderboard, suggesting an unannounced model is being tested.

zai-org created new HuggingFace repos for 'GLM-5.3-Flash' and 'GLM-5.3-Flash-BF16', potentially indicating new model releases.

OpenRouter API now lists a new model ID: z-ai/glm-5.3-flash, suggesting a potential new model from Z-AI.

Z.ai confirmed its Ox Alpha model is part of the GLM-series and that model weights will be released.

An article discussing Andreessen Horowitz's investments and a "bleak future" is gaining significant traction on Hacker News.