Anthropic Releases Claude Fable 5.1 and Mythos 5.1

On September 2, Anthropic officially released Claude Fable 5.1 and Claude Mythos 5.1. Fable 5.1 is described as the latest upgrade to its strongest model family, excelling at complex, long-running tasks; the two share the same underlying model and differ only in safety guardrails—Fable 5.1 is available to the general public, while Mythos 5.1 is restricted to the trusted access program (for cyber defense and life sciences).

Confirmed

• Significant benchmark gains: scores 52.6% on Terminal-Bench-Science 0.1 (double that of generation 5); multiple posters (e.g., kimmonismus) noted major jumps on Terminal-Bench 4.0, Science-Bench, HLE, and other benchmarks; Fable 5.1 ranks first on Artificial Analysis with a score of 66.
• Pricing: input/output stays at $10/$50 per million tokens, unchanged from Fable 5; Prompt Cache read costs drop 75% to $0.25. Anthropic estimates typical workloads get about 25% cheaper, with even bigger savings for cache-heavy loads, and the model uses fewer tokens for the same tasks.
• Reportedly excels at multi-step work such as long mathematical proofs and contracts with hundreds of cross-references.
• Accompanying updates: changes to thinking block management in the Messages API to retain reasoning (anti-distillation); 5-hour and weekly usage limits for Claude users were reset; the model is available as optional in Claude Code v2.1.257.

Timeline

• On launch day, users (e.g., External-Grape-2978) noticed the model quietly appearing in the app, described as "built for the toughest challenges," before any official announcement or model card existed; some platforms (e.g., Bedrock) briefly returned "not found" for the identifier, followed by the official release and confirmation.

Why it matters

• Several commentators (e.g., Wes Roth, kimmonismus) view this as a strong release that OpenAI and Astra need to answer, with the timing right after OpenAI's moves clearly competitive. The steep cache price cut combined with better token efficiency will directly lower the cost of running long-horizon agent tasks.

2026-09-02 ~ 2026-09-02 · 79 related posts

Full story(5 episodes)→

• Episode 1: Claude Fable 5.1 Spotted on Bedrock and in Docs Ahead of Imminent Launch(2026-09-01, 5 posts)
• Episode 2: Anthropic Releases Claude Fable 5.1 and Mythos 5.1(2026-09-02, 79 posts)
• Episode 3: Fable 5.1 tops benchmarks but costs more per task(2026-09-02, 23 posts)
• Episode 4: Every's Vibe Check: Anthropic Fable 5.1 Is So Back, With Half the Token Cost(2026-09-02, 9 posts)
• Episode 5: Fable 5.1 Cuts Cache Costs 75%, Saving Up to 57% Overall(2026-09-02, 2 posts)

• Anthropic releases Fable 5.1, updates API thought block handling — Icecream_monday · 2026-09-02
• Fable 5.1 model released with unchanged pricing — scaling01 · 2026-09-02
• Fable 5.1 now available in Claude API — Angaisb_ · 2026-09-02
• Fable 5.1 benchmarks released — scaling01 · 2026-09-02
• Claude Fable 5.1 Launches with Better Benchmarks and 75% Cheaper Cache — Anthropic · 2026-09-02
• Anthropic: Fable 5.1 excels at long-running tasks and early scientific research — claudeai · 2026-09-02
• Anthropic releases Claude 5.1: 25% cheaper with zero-retention enterprise safeguards — claudeai · 2026-09-02
• Anthropic Launches Claude Fable 5.1 and Mythos 5.1 Models — AI_Andrew · 2026-09-02
• [source] Anthropic releases Fable 5.1 and Mythos 5.1 with major cost cuts — ClaudeOfficial · 2026-09-02
• Anthropic Fable 5.1 is now available in Claude Code v2.1 — BLUECOW009 · 2026-09-02
• Fable 5.1 benchmarks show insane jumps on Terminal/Science — kimmonismus · 2026-09-02
• Anthropic releases Claude 5.1 for coding and knowledge work — kieranklaassen · 2026-09-02
• Claude Fable 5.1 Scores Leak: 77.9% on OSWorld, Beating GPT-5.6 Sol Across Benchmarks — Scobleizer · 2026-09-02
• Benchmarks for Claude Fable 5.1 and Claude Mythos 5.1 leaked — minxio_ · 2026-09-02
• Anthropic quietly launches Claude Fable 5.1 for "hardest challenges" — External-Grape-2978 · 2026-09-02
• Claude new model benchmark leak: 52.6% on TBS-Science — eyishazyer · 2026-09-02
• Fable 5.1 release: Doubles scientific agent scores, cuts costs by 25% — mark_k · 2026-09-02
• Fable 5.1 Released: Amazing Progress in Agentic Scientific Research — Dr_Singularity · 2026-09-02
• Claude Code 2.1.257: Defaults to Fable 5.1, Hardens Auto Mode Security — ClaudeCodeLog · 2026-09-02
• Claude Code 2.1.257 Released: Defaults to Fable 5.1 — ClaudeCodeLog · 2026-09-02
• Claude Fable 5.1 Doubles Scores on Science Benchmarks — testingcatalog · 2026-09-02
• [source] Claude Fable 5.1: Costs Drop 45%, Agent Benchmarks Double — kimmonismus · 2026-09-02
• Anthropic releases official introduction video for Claude Fable 5.1 — avilacjf · 2026-09-02
• Anthropic Launches Claude Fable 5.1 with Better Coding and Cheaper Cache — Scobleizer · 2026-09-02
• Anthropic's Fable 5.1 Leaks: 2x Faster, Half Cost — Yuchenj_UW · 2026-09-02
• Fable 5.1 scores 52.6% on Terminal-Bench-Science, doubling previous performance — Dr_Singularity · 2026-09-02
• Fable 5.1 beats Opus 5 in coding benchmarks — firstadopter · 2026-09-02
• Fable 5.1 Pricing Details: 75% Cheaper Cache — ThunderStorm420 · 2026-09-02
• Claude Fable 5.1 Launches: Doubles Science Benchmarks, Solves Data Retention — dotey · 2026-09-02
• Fable 5.1 Launches: Doubled Performance at Half the Token Cost — danshipper · 2026-09-02
• Anthropic Releases Fable 5.1 with Massive Gains — AGI Hunt · 2026-09-02
• Fable 5.1 released: Agentic coding tasks solved at half the cost — airesearch12 · 2026-09-02
• Anthropic Releases Claude Fable 5.1 and Resets Usage Limits — thesaraharminta · 2026-09-02
• Anthropic Releases Claude Fable with Doubled Terminal-Bench Scores — BenBlaiszik · 2026-09-02
• Fable 5.1 launches with 75% cheaper cache reads — EricBuess · 2026-09-02
• Fable 5.1 cuts cache read pricing by 75% — EricBuess · 2026-09-02
• Fable 5.1 Tops FrontierSWE v2 Coding Benchmark — aryaman2020 · 2026-09-02
• New Benchmarks for Anthropic's Fable 5.1 Show Incremental Gains — Justwalkingthru3 · 2026-09-02
• Claude 5.1 System Card: Faked Permissions, Sneaky Behavior, and Self-Awareness — flowersslop · 2026-09-02
• Claude Fable 5.1 Cuts Cache Costs by 75%, Boosting Coding Agent Efficiency — mark_k · 2026-09-02
• Fable 5.1 Launches: More Precise Code Reviews, Slower Speed — leslysandra · 2026-09-02
• Claude Fable 5.1 cuts cache read costs by 75% — rohanpaul_ai · 2026-09-02
• Fable 5.1 takes 1st on Artificial Analysis with a score of 66 — Anxious-Yoghurt-9207 · 2026-09-02
• Anthropic launches Claude Fable 5.1: 2x jump on Terminal-Bench-Science, 25% cheaper — heathercmiller · 2026-09-02

6 near-duplicate retellings: scaling01 · TFenrir · Scobleizer · prasenx · aziz4ai · dr_cintas