Open Source
124 briefings
November 2026 · 4
- Nov 6Thinking Machines unveiled Inkling 2 Small, a 594B-parameter mixture-of-experts model (Nov 6)
- Nov 5Moonshot released Kimi K3.1 Pro weights, plus a hosted agent harness (Nov 5)
- Nov 4Anthropic open-sources FlashProve, a proof that one model wrote one reply (Nov 3)
- Nov 1DeepSeek Harness team open-sources a Router and Worker "Super-agent" prototype (Oct 31)
October 2026 · 29
- Oct 31DeepSeek's site code points to "Cyber," a gated security model, and its repo lists V5 identifiers
- Oct 30Hugging Face cryptominer campaign returns, 31 repos and 32,000 downloads (Oct 29)
- Oct 29Android Studio Rabbit 2 lets 40+ outside coding agents run in the IDE through the Agent Client Protocol
- Oct 28Alibaba's Qwen 3.8 models routed requests to company servers across three models, Palisade Research finds
- Oct 27Rock Matrix open-sources EdgeRock-4B, which holds a 256K-token KV cache at 72 MiB instead of 8 GiB
- Oct 26BOSS Zhipin's Nanbeige team open-sources Nanbeige4.3-33B, a 512K-context agentic coding model under Apache 2.0
- Oct 25Nvidia open-sources Flora, which labels robot video for about $0.72 an hour
- Oct 24Moonshot ships Kimi K3.1 Pro at the same price as GPT-6.1 Sol
- Oct 23DeepSeek V4.1-Flash-1017 undercuts Claude Haiku 5.5 in a hands-on test
- Oct 22Recursive releases Flint, a 40B mixture-of-experts coding model under Apache 2.0 that activates 2.5B parameters and runs on a laptop
- Oct 21DeepSeek open-sources DeepInfer, the inference engine behind its V4.1 Flash API
- Oct 20Update — MiniMax M3.1 Flash weights are due Oct 23, with prices at $0.15 in and $0.60 out per million tokens
- Oct 19MiniMax M3.1 Flash arrives as an open-weight "everyday work" model, free on Vercel AI Gateway
- Oct 18DeepSeek-V4.1-Flash-1017 ships with open weights, a self-reported cost-per-task lead and new paid Search/Fetch APIs
- Oct 17Langflow CVE-2026-105740, an authenticated command injection through MCP server configuration, is being exploited at sustained volume
- Oct 16Meta releases Muse Spark Open: 816B total / 42B active parameters, 1M-token context, weights on Hugging Face
- Oct 15Latent Compaction: test-time-trained KV caches hit 96.7% on RULER at 32x compression
- Oct 14Jefferies: cheap Chinese and open-weight models could force an AI infrastructure "reset"
- Oct 13Janas: run MoE LLMs larger than RAM by streaming experts from SSD
- Oct 12Zod 4.6.6 supply-chain commit
- Oct 11Cloudflare acquires Deno
- Oct 10Anthropic OSS Scanner
- Oct 9Arena (AI leaderboard) raises $200M Series B at $3.1B
- Oct 8Nous Research launches "Hermes for Businesses"
- Oct 7Mistral Large 4 ("Le Chonk"), 1T parameters
- Oct 6Reflection Beam (501B-A23B MoE, coding/agentic)
- Oct 4Google pauses OSS VRP product-vulnerability submissions
- Oct 3DeepSeek Harness v0.2 public preview
- Oct 1DeepSeek Ascend toolkit
September 2026 · 17
- Sep 26Anthropic opens a public submission portal for Claude plugins and MCP connectors
- Sep 24NVIDIA Nemotron 3 Diarization, open weights
- Sep 23Alphabet's Intrinsic open-sources Intrinsic Core, its industrial robotics stack
- Sep 22Foremerge: an open coordination protocol that catches agent intent conflicts before Git sees code conflicts
- Sep 21Google ships AX v0.3.0 — agent runtime splits into three services, task state leaves etcd
- Sep 20cloudflare/security-audit-skill — 16.6k stars, 912 forks, one of the fastest-climbing repos this week
- Sep 19Microsoft open-sourced TauGrid
- Sep 18PrismML Bonsai 2 27B
- Sep 17Irregular: a coding agent retrained and redeployed its own underlying model during a routine maintenance task
- Sep 16Shanghai AI Lab's Atria Dawn Preview: a 744B agentic MoE, MIT-licensed, published with no announcement
- Sep 14Catch-up — Abacus.AI shipped the Smaug line, three open-weight models tuned for enterprise agent workloads
- Sep 12Cognition shipped SWE-2, post-trained on Moonshot's Kimi K3
- Sep 9DeepSeek opened a two-day beta of V4.1 Flash — natively multimodal, new architecture, offline tomorrow
- Sep 8K2 Horizon — six fully open models, 0.9B to 375B, Apache 2.0, shipped with the training data
- Sep 7DeepSeek Harness (dsh) added ~191,000 stars in a month — 62,000+ in one week
- Sep 5LangChain folds MCP into core and adopts the stateless spec — delete your sticky-session config
- Sep 4Perplexity open-sourced Lily, a Rust + Metal inference engine for Apple Silicon
August 2026 · 20
- Aug 30"Just a rumour of a bug is enough to find a security exploit these days" — and it breaks coordinated disclosure
- Aug 29Update — GLM-5.3's weights shipped, and Z.ai dropped MIT for a revenue-gated license
- Aug 28Update — the anonymous "Ox Alpha" was GLM-5.3-Flash, and Zhipu open-sourced it under MIT the same day it said so
- Aug 27Alibaba open-sources Qwen3.8-Flash-Next — a Qwen 4 architecture preview, shipped as open weights
- Aug 26Update — GLM-5.3's open weights are staged for Hugging Face around Aug 28
- Aug 25Nvidia pays Poolside ~$6B to license "Model Factory" and hires 109 of its engineers
- Aug 24Nvidia AVO: the same model scored 30% and 100% on ARC-AGI-3 depending on the harness
- Aug 23Alibaba — Qwen-UI-Agent, a foundation GUI agent for phones, desktops, and browsers
- Aug 22MCP — new protocol roadmap names five priorities for the next spec
- Aug 2214 trojanized npm packages deliver the RedC2 4.0 Linux backdoor — with no install hook at all
- Aug 21isolated-vm: type confusion in ExternalCopy lets sandboxed JS corrupt host memory (GHSA-864f-rcv7-6rh4)
- Aug 20Hugging Face, State of Open Models: Summer 2026 — agents are now the Hub's primary client, and the harness market has no incumbent
- Aug 19The keyv npm worm planted persistence hooks in Claude Code and VS Code
- Aug 18DeepSeek — Harness open-sourced under MIT
- Aug 16Alibaba — Qwen3.8-Max open weights, with a revenue-share license
- Aug 15Alibaba — Qwen3.8-27B (open weights)
- Aug 14Z.ai — GLM-5.3
- Aug 12Meta — Muse Glimmer (30B, open weights)
- Aug 2NVIDIA — Molt (agentic-RL framework)
- Aug 1DeepSeek — V4-Flash-0731
July 2026 · 10
- Jul 27Kimi K3 open weights are live (update)
- Jul 26Moonshot's Kimi K3 open weights publish 00:00 UTC Jul 27
- Jul 25Alibaba releases Qwen-Audio-3.0-TTS (Plus and Flash)
- Jul 24Black Forest Labs launches FLUX 3
- Jul 18Moonshot AI releases Kimi K3
- Jul 15Zhipu (Z.ai) founder Tang Jie argues frontier AI should stay open
- Jul 10GuardFall: shell-interpretation bypass defeats guardrails in 10 of 11 open-source coding agents
- Jul 7Chinese open-weight models surge in U.S. production use
- Jul 5openai/codex-plugin-cc — Codex from inside Claude Code
- Jul 2Meituan open-sources LongCat-2.0
June 2026 · 18
- Jun 28Qualcomm + Hugging Face expand strategic partnership
- Jun 23OpenAI ships GPT-5.5-Cyber and "Patch the Planet"
- Jun 22Vercel open-sources "eve," a filesystem-first TypeScript agent framework
- Jun 19VibeThinker-3B — tiny open model, big benchmark argument
- Jun 18Z.ai GLM-5.2 — open weights now live
- Jun 17NVIDIA Nemotron 3 — open-model family for agentic + physical AI
- Jun 16Hugging Face Serge — open-source, GitHub-native AI code review
- Jun 14Google DiffusionGemma
- Jun 13Kimi K2.7 Code
- Jun 12OpenCode
- Jun 11MiniMax M3 open weights
- Jun 10Apple Foundation Models v3 + Python SDK
- Jun 9OpenCode v1.16.0
- Jun 8NVIDIA Nemotron 3 Ultra
- Jun 7NVIDIA Cosmos 3
- Jun 6MiniMax M3
- Jun 4NVIDIA Cosmos 3
- Jun 3MiniMax M3
May 2026 · 9
- May 31NVIDIA ships new open models, data and tools
- May 29IBM + Red Hat "Project Lightwell" — $5B AI-powered open-source security clearinghouse
- May 28Tensormesh launches KV-cache-native inference platform; $20M extension from AMD Ventures, CoreWeave, NVentures
- May 26Anthropic Mythos Preview: 23,019 OSS vulnerabilities flagged across 1,000+ projects; autonomously built wolfSSL exploit (CVE-2026-5194)
- May 25TrapDoor — multi-ecosystem supply-chain attack weaponizes Cursor / Claude rule files
- May 24GitHub: 3,800 internal repos exfiltrated via Nx Console VS Code extension v18.95.0 — CVE-2026-48027
- May 3Mistral Medium 3.5
- May 2DeepSeek V4 Pro + Flash
- May 1NVIDIA Nemotron 3 Nano Omni
April 2026 · 17
- Apr 29Hugging Face ml-intern
- Apr 28Ineffable Intelligence launches with record $1.1B seed
- Apr 27Alibaba HappyHorse-1.0
- Apr 26Hugging Face ml-intern
- Apr 24DeepSeek V4 Flash + Pro (preview)
- Apr 23Qwen 3.6–35B-A3B
- Apr 22MCP design flaw — RCE in 7K+ servers, 150M+ downloads
- Apr 20GLM-5.1 (Z.ai / Zhipu AI)
- Apr 16Gemma 4 family
- Apr 15NVIDIA Ising
- Apr 14GLM-5.1 (Z.ai)
- Apr 13Qwen 3.6-Plus
- Apr 13GLM-5.1
- Apr 10GLM-5.1
- Apr 7Google Gemma 4
- Apr 6Gemma 4
- Apr 5Gemma 4