Models
Anthropic releases Claude Fable 5.2 and Mythos 5.2: output up to 30% faster and list prices cut 20% to $8 and $40 per million tokens; Fable is generally available with cyber and biology safeguards, Mythos is limited to vetted organizations
Claude Fable 5.2 and Mythos 5.2 launched November 9 with output up to 30% faster and list prices down 20% to $8 and $40 per million tokens.
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026. The company says output is up to 30% faster than the 5.1 versions, and list prices fell from $10 and $50 to $8 and $40 per million input and output tokens. The two are the same underlying model: Fable is generally available with cyber and biology safeguards, while Mythos is limited to vetted organizations. The speed and benchmark figures are Anthropic’s own.
The essentials
- Speed and price: up to 30% faster output, $8 per million input tokens and $40 per million output tokens, down from $10 and $50. VentureBeat lists cache reads at $0.20 (from $0.25), and a fast mode at $16 and $80 that is up to 2.5 times faster. The Batch API halves the standard rates.
- Who gets what: TestingCatalog says Fable is available across Claude’s platforms, AWS, Google Cloud and Microsoft Azure, as claude-fable-5-2 on the API. Vetted cybersecurity and life-sciences organizations can apply for Mythos. BigGo Finance says Mythos fast mode is limited to verified participants.
- Against rivals: Fortune says the new price is 20% below GPT-6 Astra’s $10 and $50, and 80% below its $1 cached-token rate. Neowin relays Anthropic’s table: 49.1% on FrontierCode 2.0 at extra-high effort, against 45.5% for Opus 5.5 and 42.8% for GPT-6 Astra.
- Caveats from the same coverage: Opus 5.5 stays half the price, and TestingCatalog notes Anthropic still points most coding and knowledge work to it. A French consultancy’s headline, SFEIR, says cost per task can rise when effort is pushed, though we saw only the headline.
What “20% cheaper” does and does not mean
The 20% is the list price per token. What a task costs also depends on how many tokens the model uses and at what effort setting: VentureBeat reports defaults of high effort on the API and medium in Claude apps, and SFEIR’s headline suggests heavy effort can offset the cut. We did not test cost per task. Also, “generally available” means no vetting is required, not that every plan includes it; we did not check plan limits.
What is independent and what is not
The benchmark table, the speed claim and the 30% figure come from Anthropic. A relay of ARC Prize’s verification reports 81.2% on ARC-AGI-3 in its standard harness, which we saw only through a newsletter. No independent price-per-task comparison was found. Not investment advice.
What is not established
- Independent speed measurements of the “up to 30% faster” claim: none found.
- Which Claude plans include Fable 5.2: not checked.
- Cost per completed task versus Fable 5.1: not established.
- Independent replication of the FrontierCode 2.0 and FrontierSWE results: none found.
Summary table
| Point | What is stated as of November 7, 2026 | Source |
|---|---|---|
| Release | November 9, 2026; model ID claude-fable-5-2 | TestingCatalog |
| Price | $8 input / $40 output per million tokens, from $10 / $50 | VentureBeat |
| Fast mode | $16 / $80, up to 2.5x faster; Mythos limited to verified users | BigGo Finance |
| Rival price | 20% below GPT-6 Astra’s list price | Fortune |
| Coding score | 49.1% on FrontierCode 2.0 versus 45.5% for Opus 5.5 (vendor table) | Neowin |
FAQ
Are Fable and Mythos 5.2 different models?
No. Anthropic says they are the same underlying model; Fable adds cyber and biology safeguards and Mythos is for vetted organizations.
How much did the price drop?
List prices fell 20%, from $10 and $50 to $8 and $40 per million input and output tokens.
Is it cheaper for every task?
Not necessarily. The cut is per token; the total depends on tokens used and effort settings, which we did not test.
Method and limits
Sources checked on November 10, 2026: VentureBeat, Fortune, Neowin, TestingCatalog, BigGo Finance and SFEIR (headline only), read as search extracts; we did not read Anthropic’s launch post directly. Benchmarks are vendor-reported. Not investment advice. Disclosure: OrigoTech drafts with Claude, a model made by Anthropic, the subject of this article.
Sources
- VentureBeat: Anthropic releases Claude Fable 5.2 and Mythos 5.2, pairing new, record-high benchmarks with 20% API cost reduction (verified 2026-11-08)
- Fortune: Anthropic debuts updated Claude Fable and Mythos models, priced 20% below OpenAI's GPT-6 Astra (verified 2026-11-08)
- Neowin: Anthropic's Claude Fable 5.2 beats GPT-6 Astra in coding, can even build an operating system (verified 2026-11-08)
- TestingCatalog: Anthropic launches Claude Fable 5.2 and Mythos 5.2 (verified 2026-11-08)
- BigGo Finance: Anthropic Cuts Fable 5.2 Token Prices to $8 as It Pushes Toward a Pre-Thanksgiving IPO (verified 2026-11-08)
- SFEIR: Claude Fable 5.2 et Mythos 5.2 : prix, benchmarks, coût réel (verified 2026-11-08)
Spotted an error in this article? Report it · Our corrections policy