{"id":115204,"date":"2026-09-02T01:07:22","date_gmt":"2026-09-02T01:07:22","guid":{"rendered":"https:\/\/youzum.net\/anthropic-releases-claude-fable-5-1-and-claude-mythos-5-1-52-6-on-terminal-bench-science-and-75-cheaper-cache-reads\/"},"modified":"2026-09-02T01:07:22","modified_gmt":"2026-09-02T01:07:22","slug":"anthropic-releases-claude-fable-5-1-and-claude-mythos-5-1-52-6-on-terminal-bench-science-and-75-cheaper-cache-reads","status":"publish","type":"post","link":"https:\/\/youzum.net\/zh\/anthropic-releases-claude-fable-5-1-and-claude-mythos-5-1-52-6-on-terminal-bench-science-and-75-cheaper-cache-reads\/","title":{"rendered":"Anthropic Releases Claude Fable 5.1 and Claude Mythos 5.1: 52.6% on Terminal-Bench-Science and 75% Cheaper Cache Reads"},"content":{"rendered":"

Anthropic has released Claude Fable 5.1 and Claude Mythos 5.1, three months after the Fable 5 line shipped in June 2026. The two are the same underlying model behind different safeguard layers. Fable 5.1 is generally available as claude-fable-5-1<\/code>; Mythos 5.1 stays restricted to vetted organizations. Both carry a 1M token context window and 128K max output tokens, with adaptive thinking always on. The headline capability number is 52.6% on Terminal-Bench-Science 0.1, against 24.7% for Fable 5 and 29.0% for Opus 5. The headline commercial number is a 75% cut to cache reads, from $1.00 to $0.25 per million tokens, which Anthropic measures as roughly 25% lower cost on typical workloads and up to 45% on agentic ones. Base input and output pricing is unchanged at $10 and $50 per million.<\/p>\n

Is it deployable?<\/strong><\/h2>\n

Yes, <\/strong>Claude Fable 5.1<\/a> is generally available as claude-fable-5-1<\/code> on the Claude API, Amazon Bedrock, Claude Platform on AWS, Google Cloud, and Microsoft Foundry. Claude Mythos 5.1 is not: it is restricted to vetted US organizations inside Project Glasswing<\/a>.<\/p>\n

The benchmark numbers<\/strong><\/h2>\n

On Terminal-Bench-Science 0.1, an agentic scientific research benchmark, Fable 5.1 scores 52.6% against 29.0% for Opus 5, 24.7% for Fable 5, and 22.4% for GPT-5.6 Sol. Anthropic reports a standard error of 3.5 to 4.5 points per model, so treat the margin, not the ranking, with care.<\/p>\n

On Terminal-Bench 4.0, Fable 5.1 reaches 55.8% and Mythos 5.1 reaches 60.9%. The gap between two identical models is the cost of safeguard interventions, which is an unusually honest disclosure. Elsewhere: CursorBench 3.2.0 at 73.4%, Humanity\u2019s Last Exam at 60.9% without tools and 65.0% with tools, AutomationBench at 31.4%, OSWorld 2.0 at 41.7% strict, and GDPval-AA v2 at 1853.<\/p>\n

Where the cost cut comes from<\/strong><\/h2>\n

Base input and output pricing is unchanged. Cache reads drop 75%, from $1.00 to $0.25 per million tokens, which is 0.025 times base input against 0.1 on every other Claude model. Anthropic measures roughly 25% lower cost on typical workloads and up to about 45% on context-heavy agentic ones. Batch processing is $5 and $25 per million tokens.<\/p>\n

Three breaking changes teams will hit<\/strong><\/h2>\n
    \n
  1. Forced tool use is gone<\/strong>: tool_choice<\/code> set to any<\/code> \u6216\u8005 tool<\/code> returns a 400. Use auto<\/code> with strict tool use or structured outputs instead.<\/li>\n
  2. Thinking blocks are model-bound<\/strong>: Fable 5.1 reads earlier models\u2019 thinking, but no earlier model reads its own. Router and fallback setups lose reasoning when they switch down.<\/li>\n
  3. Editing earlier turns invalidates thinking blocks<\/strong>: Injecting and deleting per-turn reminders, or rebuilding the system<\/code> \u6216\u8005 tools<\/code> array mid-conversation, now errors. The check is enforced for accounts created on or after August 31, 2026. The fixes are turn-scoped system messages and server-side context editing.<\/li>\n<\/ol>\n

    Additive changes: per-message effort, turn-scoped system messages, and thinking.display: \"updates\"<\/code> are all in beta behind headers. Content provenance is not optional, with a statistical text watermark on all output and C2PA credentials on files.<\/p>\n

    Anthropic also documents real regressions. Parallel tool calling is more variable, so agent loops may issue one call per turn where Fable 5 batched several. The model narrates less, answers from memory more often at low effort, and prefers whole-file rewrites over targeted edits.<\/p>\n

    Safeguards and science<\/strong><\/h2>\n

    Cyber safeguards now permit vulnerability discovery but not exploit development, cutting interventions in Claude Code by roughly 60% per session. Biology safeguards fire 85% less often on benign requests. Penetration testing, exploit generation, and binary-based vulnerability scanning still redirect to Opus.<\/p>\n

    On research, Mythos 5.1 designed protein binders with roughly 50% hit rate across 12 targets against a 10 to 15% norm, Fable 5.1 built a Venus elevation map<\/a> at 2 to 3 km resolution, and custom GPU kernels sped up seven open-source genomics models by up to 2.5x.<\/p>\n

    \n

    Key Takeaways<\/strong><\/h2>\n