Claude Fable 5.2 & Mythos 5.2: Prices Cut 20%
Fable 5.2 and Mythos 5.2 launch at lower token rates. Compare caching, batch, fast mode, workload costs, benchmarks and restricted Mythos access.
By AI Pricing Guru Editorial Team
The AI Pricing Guru editorial team verifies prices against official provider pages and maintains the datasets, calculators and guides. See our methodology and contact us to report a correction.
Report summary — 2026-11-09
- Fable 5.2 and Mythos 5.2 cut standard input and output rates 20% versus 5.1, with stronger results on several Anthropic-reported benchmarks.
- Opus 5.5 remains Anthropic's recommended daily driver; Mythos access remains restricted.
Dated story text, scoped to the report date above:
Launch prices: 5.2 vs 5.1 vs Opus
| USD per million tokens | Fable 5.2 | Fable 5.1 | Opus 5.5 | 5.2 vs 5.1 |
|---|---|---|---|---|
| Standard input | $8 | $10 | $4 | −20% |
| Standard output | $40 | $50 | $20 | −20% |
| Cache write, 5 minutes | $10 | $12.50 | $5 | −20% |
| Cache write, 1 hour | $16 | $20 | $8 | −20% |
| Cache hit / refresh | $0.20 | $0.25 | $0.20 | −20% |
| Batch input | $4 | $5 | $2 | −20% |
| Batch output | $20 | $25 | $10 | −20% |
| Pricing data: How we verify → | ||||
Launch-rate snapshot, checked November 9, 2026 against Anthropic’s official announcement and pricing documentation. Mythos 5.2 has the same rates as Fable 5.2; Mythos 5.1 matches Fable 5.1. These are dated launch figures, not continuously refreshed quotes. See maintained Anthropic pricing for subsequent changes and how we verify.
The full 1M context window uses standard per-token pricing. US data residency through inference_geo: "us" adds a 1.1× multiplier. Tool charges, taxes and cloud-specific terms can add costs.
Fast mode vs standard
| Fable / Mythos 5.2 mode | Input / 1M | Output / 1M | Generation speed | Batch compatible? |
|---|---|---|---|---|
| Standard | $8 | $40 | Baseline | No; Batch is a separate mode |
| Fast | $16 | $80 | Up to 2.5× standard | No |
| Batch | $4 | $20 | Asynchronous | Separate discounted route |
| Pricing data: How we verify → | ||||
Fast mode doubles token rates, not guaranteed end-to-end task speed. Cache and residency modifiers still apply. Fable fast mode is available through Claude Code, Claude Platform, Google Cloud and Microsoft Foundry; verify the exact managed route and its billing. Mythos fast mode requires a verification-program organization.
Monthly cost examples
Assume 10M input + 2M output tokens, identical token counts across models, no tools or residency premium. The cached scenario splits input into 1M five-minute cache-write tokens + 9M cache-hit tokens; hits must occur within valid cache lifetimes. It is a billing-volume example, not one oversized prompt.
| Monthly scenario | Fable 5.2 | Fable 5.1 | Opus 5.5 |
|---|---|---|---|
| Standard: 10 × input rate + 2 × output rate | $160 | $200 | $80 |
| Cached: 1 × write + 9 × hit + 2 × output | $91.80 | $114.75 | $46.80 |
| Batch: 10 × batch input + 2 × batch output | $80 | $100 | $40 |
| Fast, uncached: 10 × fast input + 2 × fast output | $320 | Not compared | $160 |
| Pricing data: How we verify → | |||
Computed from the official rate card above, including Opus fast rates of $8/$40. Actual caching depends on prefix reuse and expiry. Use the token calculator for your volumes and the Claude API guide for routing and hidden costs.
Benchmarks: gains are task-dependent
| Evaluation | 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 |
| OSWorld 2.1, partial offline set (%) | 89.3 | 83.5 | 87.2 | 79.5 | 81.5 |
| Humanity’s Last Exam, tools (%) | 74.2 | 65.6 | 67.7 | 57.2 | — |
| Remote Labor Index (%) | 41.3 | 17.9 | 21.3 | 20.8 | — |
| AA-Briefcase v1.1 (Elo) | 1785 | 1669 | 1800 | 1569 | 1611 |
Source: Anthropic’s launch table, not our independent testing. Effort settings differ: FrontierCode uses Xhigh for Fable and Medium for Gemini; Fable 5.2 RLI uses High. See the linked announcement and system card for methodology. Opus leads this group on AA-Briefcase Elo; Gemini leads AutomationBench. Compare current OpenAI pricing separately: benchmark scores alone do not establish cost per accepted task.
Who should switch from 5.1?
Existing Fable users should test claude-fable-5-2 on the same acceptance suite. At unchanged standard token counts, the bill falls 20%; output length, retries and tool use can change realized savings. Keep a rollback path until production-shaped tests pass.
When Opus 5.5 is the better buy
Anthropic still recommends Opus 5.5 for most coding and knowledge work. Fable 5.2 costs exactly twice as much per uncached standard input/output token. Escalate only when improved completion quality justifies that premium; reserve fast mode for a measured latency bottleneck.
Mythos access and Claude plans
Fable 5.2 is available on paid Claude plans and Claude Platform, AWS, Google Cloud and Microsoft Azure. API billing is separate from chat-plan usage; compare Claude subscription pricing.
Mythos 5.2 is restricted to vetted organizations through the Cyber Verification and Life Sciences Verification programs. Enterprise customers can use Mythos inside Claude Security; that is not unrestricted general-purpose API access.
Launch promotion, not a permanent allowance: eligible paid plans/seats with Fable receive one to three Fable usage resets based on subscription tenure; Pro and standard Team seats get Fable up to 50% of weekly limits during the promotion. Check the official promotion terms and current plan terms before buying for that temporary benefit.