Claude Fable

Evidence (18)

• 4Hacker News·9/19/2026research
  Brood War Bench tests AI agents in StarCraft play

  Brood War Bench is an experiment that measures how AI agents perform when controlling StarCraft: Brood War through agents rather than direct human input. The results show Codex Astra leading consistently, while Grok models performed poorly and often spent too much time reasoning instead of acting; older and newer models alike still struggled with basic RTS play beyond beginner level.

• 5The New Stack·9/10/2026research
  Fable 5.1 and Fable 5 tested on real benchmark tasks

  The article compares Anthropic’s Fable 5.1 and Fable 5 on a subset of Terminal-Bench-Science tasks using a user-like budget rather than the benchmark’s published conditions. In these runs, Fable 5.1 performed better on some tasks, but both models still failed several tests and the gap was much smaller than the headline benchmark score suggests.

• 7Hacker News·9/6/2026research
  GPT-6 Astra Outperforms Fable on Robot Arm Task

  A comparative benchmark on YAM robot arms found OpenAI's GPT-6 Astra completed a block-to-bowl manipulation task far more often than Claude Fable 5 and 5.1, while also using less time and lower estimated cost per run. On a more difficult puzzle-piece insertion task, Astra matched Fable 5.1's low success rate and stalled at the same final stage, suggesting stronger performance on one manipulation task but not a broader breakthrough in dexterous control.

• 6Hacker News·9/2/2026open_source
  Fable 5.1 World Modeling

  Fable 5.1 is an open-source project focused on world modeling, presented as a new release on GitHub. The repository suggests tools or methods for building and simulating world representations, attracting significant Hacker News discussion. It appears to target AI and agent-related experimentation rather than a consumer product.

• 4Simon Willison·9/1/2026model_release
  Claude Fable 5.1 Generates a Strong Animated Pelican

  Anthropic released Claude Fable 5.1 and highlighted improved performance on coding, knowledge work, and especially the new Terminal-Bench-Science benchmark. Simon Willison tested it on an SVG prompt and found that higher reasoning settings produced much better results, with the max setting generating the best pelican image he has seen from Anthropic models.

• 7The New Stack·9/1/2026product_launch
  Claude Fable 5.1 adds watermarking with code limitations

  Anthropic has launched Claude Fable 5.1 with a statistical watermark designed to help detect AI-generated text, using a method based on token-randomness changes rather than metadata. The watermark is more effective on natural language than code, and Anthropic is also tightening how preserved thinking blocks can be reused to reduce model distillation risk.

• 8Simon Willison·8/28/2026security
  AI agents are exposing security bugs within minutes

  Anil Madhavapeddy reports that patches for OCaml-related security issues are being probed by automated attackers within minutes of being discussed publicly. He and other maintainers say modern coding agents are now effective enough that even a hint of a bug can help them discover an exploit, creating pressure on existing open-source embargo and disclosure practices.

• 8Hacker News·8/28/2026security
  Agentic tools accelerate exploit discovery after bug leaks

  A maintainer describes a security fix for OCaml cohttp that was quickly followed by probes matching the bug pattern, suggesting automated watchers and agentic tools can identify exploit paths soon after a vulnerability becomes visible. The post argues that traditional embargo-based disclosure is losing effectiveness because modern LLM agents can infer and generate exploits from limited public clues.

• 8The New Stack·8/15/2026model_release
  Grok 4.6 matches top model performance at lower cost

  Grok 4.6 launched with performance that rivals or exceeds leading frontier models while keeping the same API price as Grok 4.5. The article argues that downloadable weights from competitors like DeepSeek and Qwen are increasingly capping what closed labs can charge, shifting competition toward routing and model selection rather than raw capability.

• 8Hacker News·8/12/2026model_release
  Grok 4.6 reaches AI index frontier at lower cost

  Grok 4.6 scores 61 on the Artificial Analysis Intelligence Index, tying the GPT-5.6 Sol max score and narrowing the gap to the top frontier models. It also shows strong agentic performance across knowledge work, banking, and terminal benchmarks while keeping pricing at $2/$6 per million input/output tokens.

• 4Simon Willison·8/9/2026framework_update
  Claude Opus 5 system prompt explains export-control handling

  This item quotes part of the Claude Opus 5 system prompt, which instructs the model to answer accurately about the temporary suspension and later restoration of access to Claude Fable 5 and Claude Mythos 5 due to U.S. export controls. It also says the model should avoid speculation, point users to Anthropic’s statement, and check for newer information when possible.

• 4Simon Willison·8/7/2026product_launch
  Codex Builds a Better Raccoon Heist Game with GPT-5.6 Sol Ultra

  Simon Willison compared the same one-shot game prompt across different coding agents and found Codex Desktop running GPT-5.6 Sol Ultra produced a more compelling Raccoon Heist game than a previous Claude Fable 5 attempt. The post also notes a visible bug in the generated game, which was later fixed with a follow-up prompt, and includes the full Codex transcript plus repository assets generated during the session.

• 7Anthropic News·8/7/2026framework_update
  Anthropic updates Fable 5 biology safeguards

  Anthropic says it has improved Claude Fable 5’s biology safeguards to reduce false positives and unnecessary fallbacks to a less capable model. The company says the change cuts biology-related fallbacks by about 85%, while maintaining stricter blocking for dual-use biology topics such as virology, toxicology, and molecular design.

• 4Simon Willison·8/5/2026product_launch
  Claude Fable 5 builds a raccoon heist game

  Simon Willison describes using Claude Fable 5 in Claude Code for web to generate a playable browser game based on an old raccoon-heist concept, with GitHub Pages used for rapid preview and iteration. The workflow combined the Claude agent with OpenAI's image generation API to fill in textures and visuals while the project was built and committed from mobile.

• 6The New Stack·7/29/2026model_release
  Claude Opus 5 vs. Fable 5: What the cheaper model trades off

  The article compares Anthropic’s Claude Opus 5 and Claude Fable 5, focusing on pricing, benchmark results, and hands-on reasoning tests. It concludes that Opus 5 is cheaper and often competitive, but Fable 5 remains the stronger general-access model overall in Anthropic’s own wording and in some evaluations.

• 6Hacker News·7/24/2026research
  Opus 5 Tops Artificial Analysis Intelligence Leaderboard

  Artificial Analysis reports Claude Opus 5 as the highest-scoring model on its Intelligence leaderboard, with Opus 5 variants ranked ahead of other leading systems such as Claude Fable 5 and GPT-5.6 Sol. The page compares models across intelligence, speed, latency, price, and context window using its multi-evaluation methodology.

• 5Hacker News·7/18/2026research
  Fable 5 Outperforms GPT-5.6 Sol on an NP-Hard Optimization Benchmark

  A benchmark comparison on an unpublished fiber-network optimization problem found Claude Fable 5 to be the strongest model overall, with more consistent results than GPT-5.6 Sol. The report also suggests the native /goal mode is not a universal performance boost: it sometimes improves results, but often changes the search path in ways that can also hurt outcomes.

• 7The New Stack·7/18/2026model_release
  Claude Fable 5 added permanently to subscription plans

  Anthropic says Claude Fable 5 will be included in all Max and Team Premium plans starting July 20, at 50% of usage limits. Pro and Team Standard users will still access it through usage credits and will receive a one-time $100 credit. The change reverses a planned move to remove the model from subscriptions, likely in response to competitive pressure and user demand.