Claude Fable 5.2 and Mythos 5.2: $8/$40 API Pricing and Access
Claude Fable 5.2 costs $8/$40 per 1M tokens, cache reads $0.20. API id claude-fable-5-2, 1M context. Mythos 5.2 needs vetted access.
Model releases·Published ·By RouterPlex·6 min
One API key for every model on RouterPlex: vendor list price, $0 top-up fee, from $5.
Create a key
On this page
Claude Fable 5.2 is Anthropic's new top model, released on 9 November 2026 with API id claude-fable-5-2. Claude Fable 5.2 API pricing is $8 per 1M input tokens and $40 per 1M output tokens, with cache reads at $0.20 per 1M. That is 20% below Fable 5.1. Claude Mythos 5.2 is the same model without the cyber and biology safeguards, and it is only open to vetted organizations.
The model is not on RouterPlex yet. Create a RouterPlex account today to reach 70+ models on one prepaid key, starting with Claude Opus 5.5 at $4 / $20. When Fable 5.2 lands, you change one model name and keep the same key.

Checked 9 November 2026: Anthropic announcement, Claude pricing, Fable 5.2 overview, RouterPlex catalog. Rates change. Read the checkout before a large purchase. Benchmark scores are Anthropic's, not RouterPlex measurements.
Claude Fable 5.2 API pricing
USD per 1M tokens, from the Claude pricing page. Mythos 5.2 has the same rates.
| Line | Fable 5.2 | Fable 5.1 | Opus 5.5 |
|---|---|---|---|
| Input | $8.00 | $10.00 | $4.00 |
| Output | $40.00 | $50.00 | $20.00 |
| Cache read | $0.20 | $0.25 | $0.20 |
| 5-minute cache write | $10.00 | $12.50 | $5.00 |
| 1-hour cache write | $16.00 | $20.00 | $8.00 |
| Batch input / output | $4 / $20 | $5 / $25 | |
| Fast mode input / output | $16 / $80 | $8 / $40 | |
| On RouterPlex | not yet | no | claude-opus-5-5 |
A turn with 40,000 input tokens and 2,000 output tokens, no cache, costs $0.40 on Fable 5.2 ($0.32 input plus $0.08 output). The same turn was $0.50 on Fable 5.1. On Opus 5.5 it is $0.20.
Fast mode is optional. Anthropic says it speeds up token generation by up to 2.5 times. It is not available with the Batch API. US-only inference (the inference_geo setting) multiplies every token line by 1.1.
Fable 5.2 specs and model ID
| Claude Fable 5.2 | |
|---|---|
| Claude API id | claude-fable-5-2 |
| Bedrock id | anthropic.claude-fable-5-2 |
| Mythos 5.2 id | claude-mythos-5-2 (vetted access) |
| Released | 9 November 2026 |
| Context / max output | 1M / 128K tokens |
| Thinking | Adaptive, always on |
| Default effort | high |
| Knowledge cutoff | June 2026 |
| Retirement | Not sooner than 9 November 2027 |
| Data retention | 30 days, no ZDR unless Anthropic approves |
The model runs on the Claude API, Amazon Web Services, Google Cloud, and Microsoft Azure from day one. In the Claude apps it is open to Pro, Max, Ultra, Team, and Enterprise plans.
Claude Mythos 5.2 access
Mythos 5.2 is the version for cyber defenders and life-science researchers. Anthropic calls it its most capable model for cybersecurity engineering. Fable and Mythos share one underlying model. Fable adds safeguards that block high-risk cyber work (exploit writing, binary bug hunting) and most research biology.
There are three ways in:
- Apply to the Cyber Verification Program.
- Apply to the Life Sciences Verification Program.
- Use Claude Security on a Claude Enterprise plan. It runs on the full model family, Mythos 5.2 included, with no separate check.
The price is the same $8 / $40 per 1M. Mythos fast mode is only for organizations inside the verification programs.
When Fable 5.2 blocks a cyber prompt, Opus 4.8 answers it instead, and Opus 5 answers a blocked biology prompt. Anthropic says the new model blocks about 55% fewer requests than Fable 5.1.
Fable 5.2 vs Opus 5.5 benchmarks
These are Anthropic's numbers at each model's highest effort, with safeguards on. RouterPlex did not run them.
| Benchmark | Fable 5.2 | Fable 5.1 | Opus 5.5 | GPT-6 Astra |
|---|---|---|---|---|
| FrontierCode 2.0 | 49.1% | 32.4% | 45.5% | 42.8% |
| FrontierSWE v2 | 69.9% | 56.3% | 62.3% | 65.5% |
| Terminal-Bench-Science 0.1 | 74.6% | 52.6% | 58.7% | 68.1% |
| OSWorld 2.1, partial | 89.3% | 83.5% | 87.2% | 79.5% |
| 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 Elo | 1785 | 1669 | 1800 | 1569 |
Opus 5.5 still wins the AA-Briefcase Elo row. Anthropic also reports cost per task. At high effort on FrontierCode 2.0, Fable 5.2 scores 45.3% against Opus 5.5's 45.5% for about half the cost per task. On Mythos 5.2, CyScenarioBench is 76.3% and CVE Rediscovery is 89.6%.
Anthropic still recommends Opus 5.5 as the starting point for most coding and knowledge work. Fable is for long, open-ended technical jobs.
What Fable 5.2 built in the launch tests
Anthropic gives three examples. About 1,000 Fable 5.2 agents formalized the proof of Fermat's Last Theorem in Lean in 30 hours. Opus 5.5 finished 26% of it in 18 hours with the same setup. The model also designed three working electronic devices with Diode, and scored 64.9% on the RoboDojo robotics test, against 27.6% for Fable 5.1.


Migrating from Fable 5.1 to Fable 5.2
The migration guide lists these changes that break old code.
- Change the model name from
claude-fable-5-1toclaude-fable-5-2. - Forced
tool_choice(anyortool) returns 400. - Prefill returns 400.
- Turning thinking off, or a manual
budget_tokens, returns 400. Use the effort setting. computer_20251124returns 400 except on Bedrock. Switch tocomputer_toolset_20260801.- Fable 5.1 cannot read Fable 5.2 thinking blocks.
- Leave out
temperature,top_p, andtop_k.
One behavior change bites often. Thinking counts toward max_tokens, so a small max_tokens can return no text at all. Give the model room.
Use Claude on RouterPlex while Fable 5.2 lands
RouterPlex bills vendor list price with 0% markup and a $0 top-up fee. The balance is prepaid from $5, stops at $0, and never goes negative. Each key can carry a hard spend limit. The OpenAI base URL is https://api.routerplex.com/v1 and the Anthropic base URL is https://api.routerplex.com.
Run the closest live Claude model today:
curl https://api.routerplex.com/v1/chat/completions \-H "Authorization: Bearer $ROUTERPLEX_API_KEY" \-H "Content-Type: application/json" \-d '{"model": "claude-opus-5-5","messages": [{"role": "user", "content": "Plan the migration of this service from Fable 5.1 to Fable 5.2."}]}'
Claude Code works through the Anthropic endpoint:
export ANTHROPIC_BASE_URL=https://api.routerplex.comexport ANTHROPIC_AUTH_TOKEN=$ROUTERPLEX_API_KEYclaude --model claude-opus-5-5
When claude-fable-5-2 shows up in the model catalog, change the model value. Nothing else in the code moves.
Where it goes wrong
- 400 on a Fable 5.2 call. Check for prefill, forced tool choice, or disabled thinking. All three are errors now.
- Empty reply. Your
max_tokensis too small for the thinking plus the answer. Raise it. - Model not found on RouterPlex. Fable 5.2 is not in the catalog yet. Use
claude-opus-5-5until it is. - 402 on RouterPlex. Your prepaid balance hit $0. Top up from $5 and the key works again.
- Mythos 5.2 refused. Your organization is not in a verification program. Apply first.
Next step
Create a RouterPlex account, run Opus 5.5 and 70+ other models on one key today, and switch to Fable 5.2 on the same key once it lands.
Common questions
Frequently asked questions
How much does the Claude Fable 5.2 API cost?
Anthropic lists Claude Fable 5.2 at $8 per 1M input tokens and $40 per 1M output tokens. Cache reads are $0.20 per 1M. A 5-minute cache write is $10 per 1M and a 1-hour cache write is $16 per 1M. The Batch API halves input and output to $4 and $20. Fast mode, which Anthropic says generates tokens up to 2.5 times faster, costs $16 input and $80 output per 1M. US-only inference adds a 1.1x multiplier. Fable 5.1 was $10 and $50, so the base price fell 20%. These are Anthropic's list prices, checked on 9 November 2026 on the Claude pricing page.
What is the Claude Fable 5.2 model ID?
The model ID is claude-fable-5-2 on the Claude API, Google Cloud, Microsoft Foundry, and Claude Platform on AWS. Amazon Bedrock uses anthropic.claude-fable-5-2. The context window is 1M tokens and max output is 128K tokens. Thinking is adaptive and always on, with a default effort of high. The knowledge cutoff is June 2026. Anthropic says it will not retire the model before 9 November 2027. All of this is from the Claude models overview, checked 9 November 2026.
How do I get access to Claude Mythos 5.2?
Mythos 5.2 is limited to vetted organizations that work in cybersecurity or life sciences. You apply through Anthropic's Cyber Verification Program or its Life Sciences Verification Program. Claude Enterprise customers can also reach Mythos 5.2 inside Claude Security without a separate application. The docs list the API model ID as claude-mythos-5-2 and the same $8 and $40 per 1M price as Fable 5.2. Anthropic says Fable and Mythos are the same underlying model. The difference is that Fable carries cyber and biology safeguards and Mythos does not. Checked 9 November 2026.
Is Claude Fable 5.2 on RouterPlex?
Not yet. Claude Fable 5.2 is not on RouterPlex yet as of 9 November 2026. When it lands it will use the same RouterPlex key as every other model, at vendor list price with 0% markup. Until then, the closest live Claude model is claude-opus-5-5 at $4 input and $20 output per 1M, which Anthropic itself calls its strongest daily driver. You can create an account now, top up from $5 with a $0 top-up fee, and switch the model name to Fable 5.2 once it appears in the RouterPlex catalog.
Is Claude Fable 5.2 better than Opus 5.5?
On Anthropic's own tests, Fable 5.2 scores higher on most rows. Examples: FrontierSWE v2 at 69.9% versus 62.3%, and Humanity's Last Exam with tools at 74.2% versus 67.7%. Opus 5.5 still wins the AA-Briefcase Elo, 1800 to 1785. Price is the other side. Fable 5.2 costs twice as much per token ($8 and $40 versus $4 and $20). Anthropic recommends starting with Opus 5.5 for most coding and knowledge work and moving up to Fable for long, open-ended technical jobs. All scores are vendor numbers from the 9 November 2026 announcement, not RouterPlex tests.
What changed from Fable 5.1 to Fable 5.2?
The price dropped from $10 and $50 to $8 and $40 per 1M tokens, and cache reads from $0.25 to $0.20. Anthropic says outputs are up to 30% faster. The migration guide lists breaking changes. Forced tool choice, prefill, and disabling thinking all return a 400 error. The old computer_20251124 tool returns 400 on every platform except Bedrock, so use computer_toolset_20260801. Fable 5.1 cannot read Fable 5.2 thinking blocks. Fable 5.2 also has its own rate limit, separate from the shared Fable 5.1 and Fable 5 limit. Checked 9 November 2026.
Start from a model that is listed.
This release is not on RouterPlex yet. Open a live catalog row, or browse every listed model and its price.




