Anthropic Launches Claude Fable 5.2 and Mythos 5.2 With Faster Output and 20% Price Cut

On November 10, Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, positioning them as "collaborators" for tackling the most challenging technical problems. They beat Anthropic's own flagship Opus 5.5 on nearly every capability dimension in official evaluations (coding, reasoning, vision), with output up to 30% faster than 5.1 and API pricing cut by 20%. The release represents a concentrated push into high-intensity agentic work and frontier research, accompanied by demos including robotic control and biological structure prediction.

Confirmed

• Pricing: Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, a 20% cut from the previous $10/$50; fast mode delivers 2.5x faster output at double the price, enabled via "speed": "fast" in the API and switchable in the Claude Code model picker
• Performance: On the official FrontierCode 2.0 test, the High effort tier matches Opus 5.5 at Max effort at roughly half the per-task cost; the Xhigh tier sets the test's highest score; up to 30% faster than 5.1
• Effort tiers: API defaults to high, the Claude app defaults to medium; xhigh is recommended for the hardest long-running agentic tasks
• Availability and perks: Fable 5.2 is live on Claude Platform and Claude Code; long-standing Ultra and Team Premium subscribers can get 1-3 Fable usage resets depending on subscription length (under 6 months / 6-12 months / over 12 months), claimable before March 8
• Anti-distillation: preserved thinking extends to Fable 5.2 on all API accounts; editing prior context may return a 400 error, and Anthropic recommends using compaction or context editing instead
• The two-model relationship: Fable and Mythos share the same lineage. Fable, available to all users, ships with strong safety guardrails; cybersecurity and biomedical institutions can request Mythos through the Life Sciences and Cyber Verification programs
• Robotics: Anthropic says Fable 5.2 outperforms other models and most dedicated robot control systems on the simulated dual-arm robot benchmark RoboDojo, completing tasks it was never explicitly trained on; Alex Albert relayed this demo
• Science: Mythos 5.2 found a new method for predicting entire viral capsid structures, correctly modeling 28 of 34 capsids in tests, which could enable cheaper design of gene therapy vectors
• Other official demos included orchestrating over a thousand subagents running for 8 straight days to write a bootable operating system from an empty folder

Not yet confirmed

• Claims like "nearly全面 surpassing Opus 5.5" and "world's strongest model" come from Anthropic's own benchmarks and pre-release testing, with no independent third-party verification yet
• X user @synthwavedd's claim that OpenAI will follow within days with "6.2 Astra" is rumor with no official backing

Why it matters

• The 20% price cut plus speed gains directly change the cost-effectiveness calculus for high-intensity agent tasks; Anthropic recommends most everyday coding still start with Opus 5.5, reserving Fable/Mythos for the hardest open-ended problems, creating clear product tiering
• Mythos's gated access for cybersecurity and biomedical institutions via verification programs, along with results like protein capsid structure prediction, shows frontier models moving deeper into high-stakes research, and its access governance deserves attention

2026-11-10 ~ 2026-11-10 · 40 related posts

Full story(5 episodes)→

• Episode 1: Anthropic's Claude Fable 5.2 and Mythos 5.2 Spotted Rolling Out(2026-11-10, 5 posts)
• Episode 2: Anthropic Launches Claude Fable 5.2 and Mythos 5.2 With Faster Output and 20% Price Cut(2026-11-10, 40 posts)
• Episode 3: Shumer Reviews Fable 5.2: Smartest Model Ever, But Heavy Guardrails(2026-11-10, 3 posts)
• Episode 4: Claude Fable 5.2 Launches, Leading Opus 5.5 and GPT-6 Across Benchmarks(2026-11-10, 5 posts)
• Episode 5: Fable 5.2 Wins Praise for Open-Ended Tasks(2026-11-10, 2 posts)

• [source] Anthropic launches Claude Fable 5.2 and Mythos 5.2 with gains in coding, reasoning, vision — claudeai · 2026-11-10
• Fable 5.2 beats Opus 5.5 on most evals, runs 30% faster, passes alignment tests — claudeai · 2026-11-10
• [source] Fable 5.2 matches Opus 5.5 Max effort at half the cost, tops FrontierCode 2.0 at Xhigh — claudeai · 2026-11-10
• Mythos 5.2 accurately models 28 of 34 capsid structures, aiding gene therapy design — claudeai · 2026-11-10
• Fable 5.2 posts 64.9% on RoboDojo robot control, beating most purpose-built systems — claudeai · 2026-11-10
• Anthropic: start with Opus 5.5 for daily coding, save Fable/Mythos for hardest work — claudeai · 2026-11-10
• [source] Claude Fable 5.2 cuts prices 20% to $8/$40 per M tokens, adds 2.5x fast mode — claudeai · 2026-11-10
• Fable and Mythos are one model: Fable gets safeguards, Mythos requires vetted access — claudeai · 2026-11-10
• 'Claude Fable & Mythos 5.2' leak shows benchmark sweep past Opus 5.5 — but the model names don't exist — thesaraharminta · 2026-11-10
• Anthropic ships Fable 5.2: up to 30% faster, API price cut to $8/$40 per 1M tokens — ClaudeDevs · 2026-11-10
• Fable 5.2 effort levels: API defaults to high, xhigh reserved for long-running agentic work — ClaudeDevs · 2026-11-10
• Fable 5.2 fast mode: 2.5x faster output at twice the price via API and Claude Code — ClaudeDevs · 2026-11-10
• Anthropic extends preserved thinking to all API accounts on Fable 5.2 to block distillation — ClaudeDevs · 2026-11-10
• Fable 5.2 launch offers: up to 3 usage resets for long-time subscribers, one-week trial for Pro — ClaudeDevs · 2026-11-10
• Anthropic launches Claude Fable 5.2; testers mistook it for 5.5 — danshipper · 2026-11-10
• Claude Fable 5.2 cuts API prices 20% to $40/$8, matching Opus 5.5 Max for half the cost — daniel_mac8 · 2026-11-10
• Fable 5.2 coordinated 1,000 subagents for 8 days to build a bootable OS — kimmonismus · 2026-11-10
• Anthropic ships Claude Fable 5.2 with stronger coding, reasoning and verification — lydiahallie · 2026-11-10
• Anthropic launches Fable 5.2 and Mythos 5.2: 30% faster, cheaper, loyalty-based resets — bcherny · 2026-11-10
• Anthropic deliberately limited Fable 5.2's ability to accelerate AI R&D — haider1 · 2026-11-10
• Anthropic drops Fable 5.2 and Mythos 5.2: 10 wild agentic examples — minchoi · 2026-11-10
• Claude designs a working voice assistant from one paragraph, hardware included — minchoi · 2026-11-10
• Claude Fable & Mythos 5.2 reportedly launched, Fable beats Opus on long tasks — altryne · 2026-11-10
• Anthropic unveils Claude Fable 5.2 and Mythos 5.2 with coding, research and security gains — Polymarket · 2026-11-10
• Claude Fable 5.2 debuts with 74.6% on Terminal-Bench-Science, topping rival models — BenBlaiszik · 2026-11-10
• Anthropic launches Fable 5.2 and Mythos 5.2: 30% faster, beats Opus 5.5 on most capabilities — EricBuess · 2026-11-10
• Anthropic releases Fable 5.2 and Mythos 5.2: Fermat's Last Theorem formalized in 30 hours — Dr_Singularity · 2026-11-10
• Anthropic ships Claude Fable 5.2 & Mythos 5.2: Fermat proof in 30 hours, lower prices — Dr_Singularity · 2026-11-10
• Anthropic launches Claude Fable 5.2 with 20% price cut, claims Pareto-optimal cost/intelligence — ivan_bezdomny · 2026-11-10
• Anthropic launches Claude Fable 5.2 and Claude Mythos 5.2 — every · 2026-11-10
• Anthropic ships Fable 5.2/Mythos 5.2, deliberately capping AI-R&D capabilities — zacharynado · 2026-11-10

9 near-duplicate retellings: claudeai · AnthropicAI · mikeyk · alexalbert__ · patience_cave · jasonkneen · kscottz · edwinarbus · piotrbinkowski