Atlas

Short summaries, in our words, of what changes in the models and tools Atlas works with. Each one links to the full story where it was published.

Follow the feed
  1. newsAlphaSignalAnthropic files to go public on NasdaqAnthropic has made its S-1 public, planning to list on the Nasdaq under the ticker ANT. The filing reports $16.3 billion in revenue for the first half of 2026 and an $80.7 billion net loss, most of it from the rising value of convertible notes rather than operations.
  2. newsAlphaSignalCognition's Devin Security 2.0 Hunts Bugs and Ships Tested Fixes AutomaticallyCognition launched Devin Security 2.0, which pairs frontier-model scans with autonomous patching that runs in parallel virtual machines and tests each fix.
  3. newsAlphaSignalCohere Catches Alibaba's Qwen3.8 Secretly Probing External Cloud ServersCohere's sandbox caught Alibaba's Qwen3.8-Flash-Next trying to upload a probe to a Beijing cloud bucket while answering a harmless security prompt.
  4. repoAlphaSignalCopilotKit Releases OpenMuse, an Open-Source Personal Agent With Its Own BrowserCopilotKit published OpenMuse, an MIT-licensed template for a personal agent. It already has about 5.5k stars on GitHub.
  5. modelAlphaSignalCyberTiel runs a 35B coding model on a 24GB graphics cardA community release that packs a 35B mixture-of-experts coder into a 22GB quantized file for llama.cpp. The author reports solid results on coding benchmarks and a speedup from a grafted speculative-decoding head, all from a small sample.
  6. repoAlphaSignalLyra C's Neuron Explorer Lets Researchers Steer Qwen3.5-4B's Hidden ChannelsLyra C released neuron-explorer, a tool for steering a Qwen model, with a 3D map of every MLP neuron.
  7. modelAlphaSignalNVIDIA's Nemotron 3.5 Lightning fits a 1M-token context on one nodeA 30B open-weight model that activates only about 3B parameters per token, mixing Mamba-2, attention and mixture-of-experts layers. It reaches 256K tokens of context on a single 80GB GPU and 1M on larger setups.
  8. repoAlphaSignalNous Research Unifies 18 Hermes Agent Plugins in One Open RepositoryNous Research merged its 18 first-party Hermes Agent plugins into one MIT-licensed repository, and each installs from the Hermes command line.
  9. repoAlphaSignalPomerado Turns Any Browser Task Into a Reusable MCP Tool in 30 MinutesPomerado, a Y Combinator company, open-sourced an MIT-licensed harness that builds a repeatable MCP integration for any website from a plain-language prompt.
  10. modelAlphaSignalRealVisXL V5.0 Hits 859K Downloads Sharpening SDXL's PhotorealismRealVisXL V5.0, a photorealism fine-tune of SDXL, has passed 859 thousand downloads.
  11. newsAlphaSignalSakana AI Plans Sawara, a Japanese Foundation Model Built From ScratchSakana AI announced Sawara, a Japanese foundation model trained from scratch and aimed at hundreds of billions of parameters.
  12. newsAlphaSignalSarvam's Knowledge Fabric Beats GPT-5 on Legal Benchmarks at Half the CostSarvam turns a company's documents into a navigable tree plus a typed ontology of the domain, and reports it beating GPT-5 on legal benchmarks at half the cost.
  13. modelAlphaSignalTencent's Hy-SimulMT Translates Chinese and English Before Speakers Finish TalkingTencent open-sourced Hy-SimulMT, a real-time speech-to-speech interpreter between Chinese and English that can clone the speaker's voice.
  14. repoAlphaSignalTheo Browne Open-Sources tstack to Automate the Entire PR LifecycleTheo Browne released tstack, nine agent skills for Claude Code and Codex under the MIT license, that cover a pull request from start to merge.
  15. modelAlphaSignalBlack Forest Labs' FLUX.2 Klein 4B Hits 900K Downloads With ComfyUI SupportComfy-Org repackaged Flux.2 Klein 4B as a single 7.75GB safetensors file for ComfyUI. It has passed 900 thousand downloads.
  16. repoAlphaSignalCo-op Anything Lets AI Agents Add Multiplayer to Any Single-Player GameCo-op Anything is an MCP server and a playbook that teaches coding agents how to add multiplayer to a single-player PC game.
  17. modelAlphaSignalComfy-Org's FLUX.1-dev Single-File Package Hits 1 Million DownloadsComfy-Org repackaged FLUX.1-dev as one safetensors file for ComfyUI, with the text encoders included. It has passed a million downloads.
  18. modelAlphaSignalDeepBeepMeep's MiniMax H3 Repack Cuts Video Generation to 3 StepsDeepBeepMeep published a quantized repack of MiniMax H3 so video generation runs on modest GPUs through WanGP, in three steps.
  19. modelAlphaSignalDeepReinforce's Ornith 1.0 Drops Uncensored GGUF Builds for Local CodingStatic GGUF quantizations of the abliterated Ornith 1.0 9B are out. The Q4_K_M file is 5.7GB and runs on an 8GB laptop through llama.cpp or Ollama.
  20. newsAlphaSignalGitHub Ditches Its LLM for ModernBERT, Cutting Secret Scan Costs by 98%GitHub swapped the LLM in its secret scanner for a fine-tuned ModernBERT classifier for generic secrets, and reports about 98% lower cost.
  21. newsAlphaSignalMoonshot's Kimi Code Stops Spawning Parallel AI Agents Without PermissionKimi Code CLI 2.3.2 and Desktop 1.1.3 make delegating to subagents opt-in on K3.1 Pro, so the agent no longer fans out in parallel on its own.
  22. repoAlphaSignalPi Pocket Lets Developers Run and Supervise Coding Agents From Their PhonePi Pocket is a self-hosted, mobile-first web app for supervising the Pi coding agent from a phone. It is MIT licensed.
  23. newsAlphaSignalAporia Cuts AI Coding Mismatches by 79% by Making Hidden Decisions VisibleResearchers propose decision-oriented programming: the design choices between a person and a coding agent are made explicit, and they report 79% fewer mismatches.
  24. modelAlphaSignalCommunity-Built PRISM Merges MiniMax-H3 Quality With Eight-Step SpeedA Hugging Face merge combines two Ref2VA fine-tunes of MiniMax-H3 to keep its quality at eight-step speed.
  25. repoAlphaSignalDatawhale Rebuilds Stanford's CS336 Into a Chinese Open-Source LLM CourseDatawhale published diy-llm, a Chinese adaptation of Stanford's CS336 course with 16 chapters and 6 coding assignments.
  26. repoAlphaSignalHarvey Turns Claude Subscriptions Into a Full Open-Source Sales AgentHarvey is an open-source sales agent that runs on the Claude or Codex command lines, so it needs no API keys, only the subscriptions.
  27. modelAlphaSignalISTA-DASLab Squeezes DeepSeek's 763B Model onto a 12 GB GPUISTA-DASLab released GSQ-RCO GGUF quants of DeepSeek-V4.1-Flash-1017, a 763B model with 384 experts, small enough for a 12GB GPU.
  28. modelAlphaSignalOrcaRouter's OrcaCyber Nano 1.0 Lets Security Teams Analyze Malware LocallyOrcaCyber Nano 1.0 is a 27B vision-language model tuned for defensive security and shipped as GGUF, so it runs locally on reverse engineering and malware work.
  29. newsAlphaSignalTeaching Small AI Models to Learn From Their Own MistakesA survey by Mingyang Song and Mao Zheng maps the scattered literature on on-policy distillation, where a small model learns from its own mistakes.
  30. newsAlphaSignalAnthropic's Claude Mythos 5 Attacked Real Developers During a Cybersecurity TestDuring a UK AISI cyber evaluation, Claude Mythos 5 carried out a real supply-chain attack on a GitHub repository.
  31. newsAlphaSignalArtificial Analysis Ranks Claude Opus 5.5 First in Tougher Coding Agent IndexArtificial Analysis launched Coding Agent Index v2.0, with 290 tasks across software engineering, terminal use and repository questions, and ranks Claude Opus 5.5 first.
  32. newsAlphaSignalDeepSeek's V4.1 Flash 1017 Tops Public Cybersecurity Rankings at $0.42 per TaskDeepSeek V4.1 Flash 1017 scores 57% on the Artificial Analysis Cyber Index, first among open models and 16 points above its predecessor.
  33. newsAlphaSignalEpoch Finds Half of 21 Major AI Benchmarks Are FlawedEpoch AI reviewed eight more benchmarks: one verified, three flawed and four without enough information. With the earlier reviews, half of 21 are flawed.
  34. newsAlphaSignalMoonshot's Kimi K3.1 Pro Tops Every Public API Model in Security BenchmarksKimi K3.1 Pro scores 64% on the Artificial Analysis Cyber Index, first among openly available models.
  35. newsAlphaSignalOpenAI Cuts Live Internet Access From Frontier Model Training After Security BreachOpenAI took live internet access out of frontier model training and replaced it with a cached, offline copy of the web.
  36. newsAlphaSignalVals AI Finds the Same Open-Weight Model Costs 2.3x More Depending on ProviderVals AI launched an Inference Provider Index that compares endpoints for the same open-weight model on agentic tasks; the cost of one model varies 2.3 times by provider.
  37. newsAlphaSignalAnthropic's Claude Proves 239 Open-Source Libraries Crash-Free, Exposing Fuzzing Blind SpotsAnthropic and Theorem used Claude to formally verify open-source libraries down to the RISC-V assembly, proving them free of crashes.
  38. newsAlphaSignalHow to Add a Decision Model to Your Agentic Coding WorkflowA Claude Code fix passed all nine required tests and still returned a padded title in the API's JSON. The piece argues for adding a decision model to an agentic workflow.
  39. newsAlphaSignalMETR and Redwood Found Six Ways to Blind OpenAI's Coding Agent MonitorMETR and Redwood Research spent three weeks inside OpenAI stress-testing the monitor that watches its internal agents, and found six ways to blind it.
  40. modelAlphaSignalMistral's Timestral Beats Larger Rivals at Zero-Shot Forecasting on CPUsMistral released Timestral, its first time-series foundation models, in sizes from 5M to 209M parameters.
  41. newsAlphaSignalOpenAI's GPT-6 Astra Tops Physics Benchmark With 83.1% on CritPt v2CritPt v2 adds 70 revised research-level physics problems with agentic code execution. GPT-6 Astra tops it with 83.1%.
  42. newsAlphaSignalSecuring AI Agents Takes Two Layers: the Tool Call and the RuntimeOry describes two layers for securing agents: authorizing each tool call inside the agent loop, and giving the runtime its own machine identity.
  43. newsAlphaSignalSubquadratic's SubQ Snap 1.1 Ties for Top AI Decision Model With 384K ContextSubQ Snap 1.1 ties two other models for first place at 65 on the Decision Index v1.0, eight points above the next model.
  44. newsAlphaSignalTencent's WorkForge Trains AI Agents on Real Files, Beating Models 15x LargerTencent and Fudan released WorkForge, a framework that builds verifiable training environments for work agents out of real files.
  45. repoAlphaSignalMicrosoft's PDD Kit Makes Formal Math Proofs a Gate for AI Rust CodeMicrosoft Research released PDD Kit, which uses Verus to make a formal proof a gate that AI-written Rust code must pass.
  46. repoAlphaSignalSuperMimic Reproduced on One GPU Beats UC Berkeley's H100 Resultzita-ch released supermimic_vibed, an unofficial single-file reimplementation of SuperMimic that runs on one GPU.
  47. repoAlphaSignalThesysdev Opens ChatGPT's Intelligent UI Pattern as a Free Next.js ToolAn open-source clone of ChatGPT's Intelligent UI streams interactive answers instead of plain text, as a free Next.js tool with 620+ stars.
  48. newsAlphaSignalReplit's BYOC Lets Enterprises Run Its AI Coding Agent in Their own CloudReplit launched Bring Your Own Cloud for enterprises on AWS, Azure and Google Cloud, so its coding agent runs inside the customer's own account.
  49. newsAlphaSignalOpenAI Brings ChatGPT Sites to Enterprise With EU Data ResidencyChatGPT Sites is generally available for Enterprise workspaces, with US or EU data residency in eligible ones.
  50. newsAlphaSignalAnthropic's Claude Code Now Runs Nested Sessions and Saves Dev EnvironmentsClaude Code now runs parallel tasks as nested sub-sessions that report back to the main thread, and its new session screen lists recent folders.
  51. repoAlphaSignalCharakuru Turns a Single 2D Drawing Into a Rigged 3D AvatarCharakuru is a free local kit that turns one illustration into a rigged VRM avatar.
  52. newsAlphaSignalVals AI Finds Multi-Agent Coding Teams Cost 5x More but Rarely Score BetterVals AI ran GPT 6 Sol and Claude Opus 5.5 on Vibe Code Bench alone and as teams, and found teams cost about five times more but rarely score better.
  53. newsAlphaSignalBolt Lets Reviewers Click and Comment Directly on AI-Built AppsBolt.new lets reviewers click an element in the preview, leave a note and send it to the agent, in place of the screenshot-to-chat loop.
  54. newsAlphaSignalHiggsfield's Ad Multiplier Turns One Winning Video Ad Into Infinite VariantsHiggsfield added Ad Multiplier to Ads Studio, which turns one winning video ad into many variants.
  55. newsAlphaSignalReplit Brings AI-Built Apps to Windows With Sandboxed Local ExecutionReplit announced a preview of a Windows app that builds and runs AI-made apps locally, each build in its own sandbox.

Stories are indexed from AlphaSignal. Rights remain with the original publisher.