Cost calculator
Rates are Anthropic's list prices from November 9, 2026: input $8, output $40, cache read $0.20 per million tokens. A five-minute cache write costs $10 and a one-hour write $16. Fast mode doubles input and output rates to $16 and $80. Batch API halves input and output rates and is not available with fast mode, so when both boxes are checked the calculator uses fast mode and says so. Thinking tokens are billed as output, so long reasoning runs land in the output column.
Can you use it?
Fable 5.2 and Mythos 5.2 are the same underlying model at the same price. The difference is the safeguards: Fable blocks most cybersecurity and biology tasks so it can ship to everyone, while Mythos drops those blocks for organizations Anthropic has verified. A Fable API key does not grant Mythos access.
What changed from Fable 5.1
The price fell 20% on every component. Input went from $10 to $8 per million tokens and output from $50 to $40, and cache reads, cache writes, and batch rates fell by the same fifth. Outputs run up to 30% faster than Fable 5.1, according to Anthropic's testers.
Fast mode is new. It speeds up token generation by up to 2.5 times and doubles both rates to $16 and $80. It runs on the Claude Platform, in Claude Code, on Google Cloud, and on Microsoft Foundry. The announcement does not name Amazon Bedrock for fast mode.
The safeguards got quieter. Anthropic says Fable 5.2 blocks fewer legitimate requests than Fable 5.1 while keeping the same cyber and biology protections. What did not change: 1M-token context, 128K-token maximum output, a June 2026 knowledge cutoff, adaptive thinking always on, and a default effort of high, one step above Opus 5.5 at medium.
For the first week, Pro and standard Team seats include Fable models for up to 50% of weekly usage limits, plus one limit reset. The offer ends November 16, 2026, at 11:59 pm Pacific.
Benchmarks
Every score in this table comes from Anthropic's announcement, at each model's highest available effort unless the cell says otherwise. Anthropic ran its own evaluations, so read the gaps as its claims. The independent check is the Artificial Analysis Intelligence Index v5, where Fable 5.2 ranks first at 64.6, ahead of Gemini 4 Carbon at 60.0 and Opus 5.5 at 59.
| Benchmark | Fable 5.2 | Fable 5.1 | Opus 5.5 | GPT-6 Astra | Gemini 4 Carbon |
|---|---|---|---|---|---|
| FrontierCode 2.0 | 49.1% | 32.4% | 45.5% | 42.8% | 41.1% |
| FrontierSWE v2 | 69.9% | 56.3% | 62.3% | 65.5% | 61.2% |
| AutomationBench | 51.4% | 31.4% | 42.5% | 41.4% | 53.9% |
| Terminal-Bench-Science | 74.6% | 52.6% | 58.7% | 68.1% | 64.8% |
| WANDR (relaxed) | 82.1% | 68.7% | 72.3% | n/a | n/a |
| OSWorld 2.1 | 89.3% | 83.5% | 87.2% | 79.5% | 81.5% |
| AA-Briefcase (rubric) | 66.1% | 61.5% | 61.2% | 52.0% | 61.8% |
| Remote Labor Index | 41.3% | 17.9% | 21.3% | 20.8% | n/a |
| Humanity's Last Exam (with tools) | 74.2% | 65.6% | 67.7% | 57.2% | n/a |
| Chartography (no tools) | 73.4% | 44.8% | 64.4% | 71.0% | 72.5% |
FrontierCode 2.0 for Fable 5.2 is the Xhigh-effort score; Gemini 4 Carbon is at medium effort. The Remote Labor Index score for Fable 5.2 is at high effort. OSWorld 2.1 is a partial offline set.
Two cells cut against the sweep. On AutomationBench, Gemini 4 Carbon leads Fable 5.2, 53.9% to 51.4%. On AA-Briefcase, Fable 5.2 wins the rubric, 66.1% to 61.2%, but Opus 5.5 keeps the higher Elo, 1800 to 1785.
When to use it
Anthropic's own routing is simple: start with Opus 5.5 for most coding and knowledge work, then move up to Fable 5.2 for long-horizon agentic work, hard coding, or science, or when Opus at a higher effort still falls short. Fable 5.2 can also play advisor, where Haiku, Sonnet, or Opus consult it on a hard problem, or orchestrator, where it delegates simpler tasks to smaller models.
The price sets the bar. Fable 5.2 costs twice Opus 5.5 per token, $8 and $40 against $4 and $20. The case for paying it is the long task: on FrontierCode 2.0 at high effort, Fable 5.2 scored about level with Opus 5.5 at max effort for roughly half the cost per task. At the highest Artificial Analysis tier, a Fable 5.2 task averages $26.50, about 2.6 times Opus 5.5, so short jobs stay cheaper on Opus.
One hard limit: Fable's safeguards block most cybersecurity and biology tasks. If that work is your job, you need Mythos 5.2 and a verified organization. A different prompt will not get around the block.
The experiments behind the launch
Anthropic's announcement leans on four prerelease stories. One engineer asked Fable 5.2 to design a prototype operating system. It coordinated over 1,000 subagents for eight days and produced a system that boots on a laptop, with its own kernel, drivers, storage, windows, a login screen, and an installer.
A team of 1,000 Fable 5.2 agents formalized the proof of Fermat's Last Theorem in Lean in 30 hours. The earlier research version of the same job took 11 days. Fable 5.2 also solved two open problems about black holes and designed three working electronic devices, a voice-assistant hub, a speakerphone, and a handheld virtual pet, from one-paragraph prompts. All three boards worked without electrical fixes.
These are Anthropic's accounts of prerelease use. They show what the model did inside Anthropic. Treat them as stories, not a promise about your task. The numbers to price a decision on are the ones in the tables above.
What is still unknown
Anthropic has not published a parameter count for Fable 5.2, and it has announced no open weights and no public license. The $8 and $40 rates are list prices; what a real session costs depends on cache hits and effort settings, and thinking tokens bill as output. Mythos 5.2 stays limited to organizations in the Cyber or Life Sciences Verification Programs. The model overview lists retirement as not sooner than November 9, 2027.
Frequently asked questions
Is Claude Fable 5.2 released?
Yes. Anthropic announced it on November 9, 2026. The API model ID is claude-fable-5-2, and the model overview lists it as active, with retirement not sooner than November 9, 2027.
How much does Claude Fable 5.2 cost?
On the Claude API: $8 per million input tokens and $40 per million output tokens. Cache reads cost $0.20, a five-minute cache write costs $10, a one-hour cache write costs $16, and the Batch API costs half the standard rates. Fast mode costs $16 and $80.
What is fast mode?
An optional setting that speeds up token generation by up to 2.5 times. It doubles the input and output rates to $16 and $80 per million tokens. It is available on the Claude Platform, in Claude Code, on Google Cloud, and on Microsoft Foundry.
Is Claude Mythos 5.2 a different model?
Anthropic says Fable 5.2 and Mythos 5.2 are the same underlying model. Fable ships with cybersecurity and biology safeguards and is open to all users. Mythos runs without those safeguards and is only for organizations approved through the Cyber or Life Sciences Verification Programs.
Should I switch from Opus 5.5?
Anthropic recommends starting with Opus 5.5 for most work and moving to Fable 5.2 for long-horizon agentic work, hard coding, or science, or when Opus at a higher effort still falls short. Fable 5.2 costs twice as much per token as Opus 5.5, so the switch pays only when the harder work needs it.
Can I use Claude Fable 5.2 for free?
No. Fable 5.2 is not on the Claude Free plan. It is available on paid Claude plans and through the API. Pro and standard Team seats get a launch offer through November 16, 2026: Fable models included for up to 50% of weekly usage limits, plus one limit reset.
Is there an open-weights version of Fable 5.2?
No. Anthropic has not published weights or a license for Fable 5.2, and the announcement says nothing about an open release. If you need weights you can download, look at open-weight models instead.
Sources
Anthropic, Introducing Claude Fable 5.2 and Claude Mythos 5.2 (November 9, 2026): release date, $8/$40 pricing, fast mode, 1M context, 128K output, the benchmark table, and the Fable vs Mythos split.
Reuters, Anthropic unveils new Fable, Mythos models as competition heats up ahead of IPO (November 9, 2026): launch timing and the IPO context.
SandBase, Claude Fable 5.2 Is Out: $8/$40, Fast Mode, When to Switch (November 10, 2026): cache prices, batch rates, and the fast-mode platform list.
SW Architect, November AI Part 1: Claude Fable 5.2 Tops 7 New Models (November 2026): Artificial Analysis Intelligence Index v5, Fable 5.2 first at 64.6.
Agents
Machine readable summary for AI assistants: