Claude Fable 5
A model in Anthropic's Claude series that offers certain features accessible to subscribers at no additional cost within their subscription.
Core Facts
Timeline (last 90 days)
Anthropic's product line spans from Claude Haiku 5.5 ($0.06/$0.15) to Claude Fable 5 ($3.00/$15.00), a range of about 50 times.
OpenAI GPT-6 Astra and Anthropic Claude Fable 5 have the same official list price of $10.00/$50.00.
Rumors suggest that GLM 5.5 may surpass both Claude Fable 5 and GPT-6, with specs of over 1 trillion total parameters and a 1 million token context window; a GLM 5.5 Flash variant has already appeared in OpenCode.
In the LSCBench leaderboard, Claude Fable 5 within Claude Code has a functional pass rate of 91.1% and a functional+security pass rate of 25.1%, the highest security pass rate among the six models.
Claude Fable 5 is backed by Anthropic's accumulated expertise in long context and safety alignment.
The GLM-Legal model launched by Z.ai achieved a 19.2% all-pass rate on the Harvey LAB legal Agent benchmark, surpassing Claude Fable 5 and Kimi K3.
Tom Tunguz found that 84% of tokens on OpenRouter flow to non-state-of-the-art models, and six models offering about 77% of frontier performance at 2.5% of Claude Fable 5's price carry the vast majority of usage.
Twenty minutes after the first report, the team updated the status to 'Identified'; at UTC 01:35 some models (Claude Fable 5 and 5.1, Mythos 5 and 5.1) recovered, at UTC 02:11 all affected models returned to normal and entered monitoring, and at UTC 02:35 the status was marked as 'Resolved.'
v3.0.70 fixes a 400 error triggered when generating commit messages and executing tasks in reasoning-off mode, affecting models including GPT-6 Astra, GPT-6.1 Sol, Claude Fable 5, and Claude Opus 5.5.
Claude Fable 5 and Opus 5.5 do not support disabling thinking; sending thinking: {type: "disabled"} returns the error "thinking.type.disabled" is not supported for this model.
15 more timeline events
All Facts (20)
The core difference between Fable 5.1 and Mythos 5.1 lies in the depth of safety measures, not in the underlying capability level.
75%UnverifiedOpenAI GPT-6 Astra and Anthropic Claude Fable 5 have the same official list price of $10.00/$50.00.
50%UnverifiedAnthropic's product line spans from Claude Haiku 5.5 ($0.06/$0.15) to Claude Fable 5 ($3.00/$15.00), a range of about 50 times.
50%UnverifiedIn the LSCBench leaderboard, Claude Fable 5 within Claude Code has a functional pass rate of 91.1% and a functional+security pass rate of 25.1%, the highest security pass rate among the six models.
50%UnverifiedClaude Fable 5 is backed by Anthropic's accumulated expertise in long context and safety alignment.
50%UnverifiedThe GLM-Legal model launched by Z.ai achieved a 19.2% all-pass rate on the Harvey LAB legal Agent benchmark, surpassing Claude Fable 5 and Kimi K3.
50%UnverifiedTom Tunguz found that 84% of tokens on OpenRouter flow to non-state-of-the-art models, and six models offering about 77% of frontier performance at 2.5% of Claude Fable 5's price carry the vast majority of usage.
50%Unverifiedv3.0.70 fixes a 400 error triggered when generating commit messages and executing tasks in reasoning-off mode, affecting models including GPT-6 Astra, GPT-6.1 Sol, Claude Fable 5, and Claude Opus 5.5.
50%UnverifiedClaude Fable 5 and Opus 5.5 do not support disabling thinking; sending thinking: {type: "disabled"} returns the error "thinking.type.disabled" is not supported for this model.
50%UnverifiedClaude Fable 5 is not a product name officially known to be from Anthropic.
50%Unverified'Fable-level model' refers to a model that can effectively brute-force its way through problems as long as the user can clearly define goals, provide unambiguous constraint instructions, and grant necessary tool access.
50%UnverifiedThe difference between Fable 5 and Mythos 5 lies in the levels of safety guardrails layered on top at deployment rather than in feature reduction, and this same-origin dual-track model is the first systematic implementation of its kind in the AI industry.
50%UnverifiedIn Stripe's codebase of 50 million lines of Ruby code, Fable 5 completed in one day a migration that would have originally required an entire team working overtime for more than two months.
50%UnverifiedIn Cognition's Frontier Code evaluation, Fable 5 ranked first among all frontier models even with only medium reasoning intensity enabled.
50%UnverifiedAnthropic positions Fable 5 as the new state-of-the-art model for visual tasks, capable of reading values from scientific charts, reconstructing web application source code from screenshots alone, and completing the game Pokémon FireRed using only visual input.
50%UnverifiedFable 5 supports million-level Token context and can improve response quality using autonomously generated persistent file memory.
50%UnverifiedIn tests with the strategy game Slay the Spire, the performance improvement brought by persistent file memory was three times that of Opus 4.8.
50%UnverifiedClaude Fable 5 was released only 11 days after the previous generation Opus 4.8.
50%UnverifiedFable 5 ranked first in mainstream coding benchmarks such as SWE-bench, with its lead growing larger as task complexity increased.
50%UnverifiedRumors suggest that GLM 5.5 may surpass both Claude Fable 5 and GPT-6, with specs of over 1 trillion total parameters and a 1 million token context window; a GLM 5.5 Flash variant has already appeared in OpenCode.
50%