Claude Fable, Claude Mythos and Claude Opus are three of Anthropic's five Claude model lines, and the newest models in each, Fable 5.2, Mythos 5.2 and Opus 5.5, were released between 22 September and 9 November 2026. This page puts the three lines side by side: API IDs, release dates, access rules, context window, prices, documented behaviour changes, provider-reported benchmarks, data retention and retirement dates. It covers the Fable and Mythos 5.x models and the three most recent Opus models (Opus 5.5, Opus 5 and Opus 4.8). It is written for developers, buyers and security teams. Lines are listed alphabetically (Fable, Mythos, Opus) and versions newest first; order is never a ranking. Every fact comes from Anthropic's own documentation and announcements, cited in the sources list.
Last checked: 2026-11-15.
This is an independent reference page from NHI Mgmt Group based on public provider documentation; it is not sponsored or endorsed by any provider. It uses Anthropic's public documentation as of November 2026. Model details change, so confirm against Anthropic's current documentation.
Key takeaways
- Fable and Mythos are one model in two configurations. Anthropic's 9 November 2026 announcement states that "Claude Fable and Claude Mythos are the same underlying model": Fable carries cybersecurity and biology safeguards and is open to all customers, Mythos is for vetted organisations.
- Opus 5.5 costs half of Fable 5.2 per token. Opus 5.5 lists $4 input and $20 output per million tokens; Fable 5.2 and Mythos 5.2 list $8 and $40.
- All three current models share 1M context and 128K output. Each has adaptive thinking that is always on; default effort is high for Fable 5.2 and Mythos 5.2 and medium for Opus 5.5.
- Retention rules differ by line. Every Fable and Mythos 5.x model is a Covered Model requiring 30-day retention; the Opus 5.5 announcement states zero data retention is available, as with previous Opus models.
- Anthropic names Opus 5.5 as the starting point. The models overview says to start with Opus 5.5 for most workloads and to use Fable 5.2 when evaluations on Opus 5.5 at higher effort still fall short.
What are Claude Fable, Mythos and Opus?
Claude Fable is Anthropic's generally available top-tier line, first released as Fable 5 on 9 June 2026. The Fable 5.2 model page lists its uses as multi-hour agent sessions, multistep deep research and analysis carried through to a finished document, spreadsheet or deck.
Claude Mythos is the restricted-access counterpart. Each numbered Mythos release ships on the same day as a Fable release; the Mythos 5.2 model page is subtitled "Claude Fable 5.2 for verified organizations" and states the same specifications and pricing. Anthropic's announcement calls Mythos 5.2 its "most capable model for cybersecurity engineering". Access requires verification through a programme such as the Cyber Verification Program.
Claude Opus is Anthropic's Opus model line. Anthropic's Fable and Mythos 5.2 announcement calls Opus 5.5 its "strongest daily driver" and names it as the starting point for most coding and knowledge tasks. The Opus 5.5 announcement claims it "performs at the level of Claude Fable 5.1 on most work".
All three lines are proprietary: Anthropic publishes no weights and states no parameter count. For the full Claude line-up, see the Anthropic Models hub; for single-model detail, see Claude Fable 5.2, Claude Mythos (Preview, 5, 5.1, 5.2) and Claude Opus 5.5.
Every model at a glance
As of 2026-11-15. Source: Anthropic models overview, model pages, model deprecations page and release notes.
| Model | API ID | Released | Status |
|---|---|---|---|
| Claude Fable 5.2 | claude-fable-5-2 | 2026-11-09 | Active (latest) |
| Claude Fable 5.1 | claude-fable-5-1 | 2026-09-01 | Active |
| Claude Fable 5 | claude-fable-5 | 2026-06-09 | Active |
| Claude Mythos 5.2 | claude-mythos-5-2 | 2026-11-09 | Active (verification required) |
| Claude Mythos 5.1 | claude-mythos-5-1 | 2026-09-01 | Active (verification required) |
| Claude Mythos 5 | claude-mythos-5 | 2026-06-09 | Active (verification required) |
| Claude Mythos Preview | claude-mythos-preview | Not stated in the release notes checked (Project Glasswing preview) | Deprecated 2026-06-09 |
| Claude Opus 5.5 | claude-opus-5-5 | 2026-09-22 | Active (latest) |
| Claude Opus 5 | claude-opus-5 | 2026-07-24 | Active |
| Claude Opus 4.8 | claude-opus-4-8 | 2026-05-28 | Active |
Every model listed takes text and image input and returns text. Open weights: none. Licence: proprietary. Parameters: not stated by Anthropic. See Claude Opus 5.5 vs Opus 5 vs Opus 4.8 for the Opus line in depth and Claude Mythos Versions Compared for every Mythos release.
When was each model released?
As of 2026-11-15. Dates are as shown in Anthropic's Claude Platform release notes, shown chronologically.
| Date | Event | API ID |
|---|---|---|
| 2026-05-28 | Claude Opus 4.8 released; effort defaults to high; non-default temperature, top_p or top_k returns a 400 error | claude-opus-4-8 |
| 2026-06-09 | Claude Fable 5 (all customers) and Claude Mythos 5 (Project Glasswing participants) released; Mythos Preview deprecated | claude-fable-5, claude-mythos-5 |
| 2026-07-01 | Access to Fable 5 and Mythos 5 restored | claude-fable-5, claude-mythos-5 |
| 2026-07-24 | Claude Opus 5 released at $5/$25, the same price as Opus 4.8 | claude-opus-5 |
| 2026-08-19 | Computer use and browser use toolsets available for Fable 5, Mythos 5, Opus 5 and Opus 4.8 on the Claude API | computer_toolset_20260801, browser_toolset_20260801 |
| 2026-09-01 | Claude Fable 5.1 and Claude Mythos 5.1 released at $10/$50; cache reads $0.25 | claude-fable-5-1, claude-mythos-5-1 |
| 2026-09-22 | Claude Opus 5.5 released at $4/$20 | claude-opus-5-5 |
| 2026-10-15 | Opus 5.5 fast mode leaves research preview; Opus 5 and Opus 4.8 fast mode remain in research preview | claude-opus-5-5 |
| 2026-10-30 | Opus 4.8 fast mode deprecated, removal 2026-11-30 | claude-opus-4-8 |
| 2026-11-09 | Claude Fable 5.2 and Claude Mythos 5.2 released at $8/$40; fast mode for Fable 5.2 on the Claude API | claude-fable-5-2, claude-mythos-5-2 |
How much does each model cost?
As of 2026-11-15. USD per million tokens, standard global rates, from Anthropic's pricing page.
| Model | Input | Cached input | Cache write | Output | Batch (input / output) | Notes stated by provider |
|---|---|---|---|---|---|---|
| Claude Fable 5.2 | $8.00 | $0.20 | $10 (5m), $16 (1h) | $40.00 | $4 / $20 | Fast mode $16 / $80 |
| Claude Fable 5.1 | $10.00 | $0.25 | $12.50 (5m), $20 (1h) | $50.00 | $5 / $25 | No fast mode price listed |
| Claude Fable 5 | $10.00 | $1.00 | $12.50 (5m), $20 (1h) | $50.00 | $5 / $25 | No fast mode price listed |
| Claude Mythos 5.2 | $8.00 | $0.20 | $10 (5m), $16 (1h) | $40.00 | $4 / $20 | Fast mode $16 / $80 (Claude API) |
| Claude Mythos 5.1 | $10.00 | $0.25 | $12.50 (5m), $20 (1h) | $50.00 | $5 / $25 | No fast mode price listed |
| Claude Mythos 5 | $10.00 | $1.00 | $12.50 (5m), $20 (1h) | $50.00 | $5 / $25 | No fast mode price listed |
| Claude Opus 5.5 | $4.00 | $0.20 | $5 (5m), $8 (1h) | $20.00 | $2 / $10 | Fast mode $8 / $40 |
| Claude Opus 5 | $5.00 | $0.50 | $6.25 (5m), $10 (1h) | $25.00 | $2.50 / $12.50 | Fast mode $10 / $50 (research preview) |
| Claude Opus 4.8 | $5.00 | $0.50 | $6.25 (5m), $10 (1h) | $25.00 | $2.50 / $12.50 | Fast mode deprecated, removal 2026-11-30 |
Cache read multipliers: the pricing page states cache reads at 0.025x base input for the Fable and Mythos 5.x models and 0.05x for Opus 5.5, so Fable 5.2, Mythos 5.2 and Opus 5.5 all list a $0.20 cache read despite different base prices.
Long context and tokeniser: Claude 4.6 and later models bill the full 1M-token window at standard rates, and fast mode prices apply across the full window. All models on this page use the newer tokeniser introduced with Claude 4.7, which the pricing page says produces about 30% more tokens for the same text than the previous tokeniser.
Data residency: setting inference_geo to "us" applies a 1.1x multiplier to all token categories for Claude 4.6 and later models on the Claude API and Anthropic-operated cloud offerings. Fast mode is not available with the Batch API. For cost guardrails, see How to set spend limits on LLM API keys and the LLM Pricing Tracker.
What changed between models?
Only changes Anthropic documents in its release notes and announcements are listed, grouped by line.
- Fable 5 to 5.1: forced tool use (tool_choice any and tool) returns a 400 error; cache reads moved to 0.025x base input ($0.25); output carries a text watermark; per-message effort changes arrived in beta.
- Fable 5.1 to 5.2: prices fell to $8/$40; the prefix check applies to every account; computer use requires computer_toolset_20260801 (computer_20251124 returns a 400 error except on Amazon Bedrock); an advisor-tool executor accepts only Fable 5.2 or Mythos 5.2 as advisor; Anthropic reports about 55% fewer classifier blocks than Fable 5.1 in early testing.
- Mythos 5 to 5.1 to 5.2: each Mythos release carries the same API changes and prices as its Fable twin. Access wording moved from "Project Glasswing participants" (5 and 5.1 release notes) to "organizations verified through Anthropic's verification programs" (5.2).
- Opus 4.8 to 5: same $5/$25 price; full effort ladder (low, medium, high, xhigh, max); disabling thinking allowed only at effort high or below.
- Opus 5 to 5.5: price fell to $4/$20; thinking can no longer be disabled (both disabled and manual enabled thinking return a 400 error); tool_choice any and tool return a 400 error; computer use requires computer_toolset_20260801 on the Claude API and Google Cloud; default effort is medium. Anthropic states Opus 5.5 runs about 40% cheaper than Opus 5 on typical workloads.
Cross-line behaviour: Fable 5.2 blocks high-risk cyber and most research biology tasks with classifiers. Blocked cyber queries are answered by Opus 4.8 and blocked biology queries by Opus 5, per the announcement. Opus 5.5 re-routes "most cybersecurity tasks" to Opus 4.8. For thinking and effort background, see Reasoning Models Explained.
What workloads does Anthropic position each model for?
This section records the provider's own positioning wording; it is not an assessment.
| Model | Positioning stated by provider | Related notes stated by provider |
|---|---|---|
| Claude Fable 5.2 | "For demanding reasoning and long-horizon agentic work" (models overview) | Use when evaluations on Opus 5.5 at higher effort still fall short |
| Claude Mythos 5.2 | "Most capable model for cybersecurity engineering" (announcement) | Verified organisations only; Claude Security runs on Mythos 5.2 for Claude Enterprise customers |
| Claude Opus 5.5 | "For long-running agentic coding and knowledge work" (models overview) | "Start with Claude Opus 5.5 for most workloads" |
| Claude Opus 5 and 4.8 | Not stated on the pages checked (2026-11-15) | Opus 4.8 and Opus 5 answer some requests blocked by Fable 5.2 classifiers |
Worked cost example
Using the list prices above, a workload of 1,000 agent runs a day, each using 200K input tokens and 20K output tokens, uses 200 million input and 20 million output tokens a day. On Opus 5.5 that is $800 input plus $400 output, $1,200 a day. On Fable 5.2 or Mythos 5.2 it is $1,600 plus $800, $2,400 a day. On Opus 5 or Opus 4.8 it is $1,500 a day, and on Fable 5.1 or Mythos 5.1 $3,000 a day. Default effort (medium on Opus 5.5, high on Fable 5.2) changes how many thinking tokens are produced. Re-run the arithmetic with token counts measured on your own data.
What is the context window of each model?
As of 2026-11-15. From Anthropic's models overview, model pages and release notes.
| Model | Context window | Max output | Knowledge cutoff | Input | Output | Reasoning controls |
|---|---|---|---|---|---|---|
| Claude Fable 5.2 | 1M tokens | 128K tokens | Jun 2026 (reliable and training) | Text, image | Text | Adaptive, always on; default effort high |
| Claude Fable 5.1, Fable 5, Mythos 5.1, Mythos 5 | 1M tokens | 128K tokens | Not stated by Anthropic on the pages checked (2026-11-15) | Text, image | Text | Adaptive, always on; disabled thinking and manual budgets return a 400 error |
| Claude Mythos 5.2 | 1M tokens | 128K tokens | Jun 2026 (reliable and training) | Text, image | Text | Adaptive, always on; default effort high |
| Claude Opus 5.5 | 1M tokens | 128K tokens (up to 300K on the Message Batches API with a beta header) | Jun 2026 (reliable and training) | Text, image | Text | Adaptive, always on; default effort medium |
| Claude Opus 5 | 1M tokens | 128K tokens | Not stated by Anthropic on the pages checked (2026-11-15) | Text, image | Text | Thinking on by default; effort low to max; disabling allowed at high or below |
| Claude Opus 4.8 | 1M tokens | 128K tokens | Not stated by Anthropic on the pages checked (2026-11-15) | Text, image | Text | Default effort high |
See Context Windows Explained.
Capabilities and tools compared
As of 2026-11-15. Cloud IDs are from the individual model pages on platform.claude.com.
| Model | Function calling and structured outputs | Computer use tool | Fine-tuning | Cloud IDs stated by provider |
|---|---|---|---|---|
| Claude Fable 5.2 | Tool use supported; tool_choice any and tool return a 400 error | computer_toolset_20260801 required (computer_20251124 accepted only on Amazon Bedrock) | Not stated by Anthropic on the pages checked (2026-11-15) | Amazon Bedrock global.anthropic.claude-fable-5-2; Google Cloud, Microsoft Foundry and Claude Platform on AWS claude-fable-5-2 |
| Claude Mythos 5.2 | Same as Fable 5.2 | Same as Fable 5.2 | Not stated by Anthropic on the pages checked (2026-11-15) | Amazon Bedrock global.anthropic.claude-mythos-5-2; Google Cloud and Microsoft Foundry claude-mythos-5-2 |
| Claude Fable 5.1 and Mythos 5.1 | tool_choice any and tool return a 400 error; strict tool use or structured outputs recommended | Toolsets available from Fable 5 and Mythos 5 (2026-08-19) | Not stated by Anthropic on the pages checked (2026-11-15) | Not stated by Anthropic on the pages checked (2026-11-15) |
| Claude Opus 5.5 | tool_choice any and tool return a 400 error; auto with strict tool use recommended; inline tools in beta | computer_toolset_20260801 on the Claude API and Google Cloud; computer_20251124 still works on Amazon Bedrock | Not stated by Anthropic on the pages checked (2026-11-15) | Amazon Bedrock anthropic.claude-opus-5-5; Google Cloud, Microsoft Foundry and Claude Platform on AWS claude-opus-5-5 |
For agent tooling risks, see Computer-Use and Browser Agents.
Where is each model available?
| Model | Availability stated by provider |
|---|---|
| Claude Fable 5.2 | Claude API, Amazon Bedrock, Claude Platform on AWS, Claude Models on Google Cloud (Google managed and Anthropic managed), Microsoft Foundry; open to all customers |
| Claude Mythos 5.2 | Same platforms as Fable 5.2; only for organisations verified through the Cyber Verification Program, the Life Sciences Verification Program or similar |
| Claude Opus 5.5 | Claude API, Amazon Bedrock, Claude Platform on AWS, Claude Models on Google Cloud (Google managed and Anthropic managed), Microsoft Foundry |
| Claude Fable 5.1, Mythos 5.1, Opus 5, Opus 4.8 | Active on the Claude API; partner cloud listings set by each cloud |
Fast mode coverage differs: Fable 5.2 on the Claude API, Google Cloud (both managed variants) and Foundry deployments hosted on Anthropic; Mythos 5.2 on the Claude API; Opus 5.5 on the Claude API, Google Cloud (Anthropic managed) and Azure-hosted Foundry deployments.
Model cards and system cards
| Model | Model card or page | System card | Technical report |
|---|---|---|---|
| Claude Fable 5.2 | Model page; Announcement | Combined Fable 5.2 and Mythos 5.2 system card, linked from the model pages | Not stated by Anthropic on the pages checked (2026-11-15) |
| Claude Mythos 5.2 | Model page; Announcement | Combined Fable 5.2 and Mythos 5.2 system card, linked from the model pages | Not stated by Anthropic on the pages checked (2026-11-15) |
| Claude Opus 5.5 | Model page; Announcement | System card | Not stated by Anthropic on the pages checked (2026-11-15) |
This page does not summarise system card contents.
Benchmarks
As of 2026-11-15. Figures are provider-reported, from Anthropic's Fable and Mythos 5.2 announcement (9 November 2026), which scores each model at its highest available effort unless noted, with safeguards on unless noted. Only rows with a figure for each model shown are listed; Anthropic published Mythos 5.2 figures for few benchmarks, so Mythos 5.2 is shown separately below.
| Model | Provider-reported benchmarks | Independent leaderboard results |
|---|---|---|
| Claude Fable 5.2 | FrontierSWE v2 69.9%; AutomationBench 51.4%; Terminal-Bench-Science 0.1 74.6%; OSWorld 2.1 (offline, partial) 89.3%; Humanity's Last Exam (with tools) 74.2% | Not checked for this page |
| Claude Mythos 5.2 | Terminal-Bench-Science 0.1 75.3%; CVE Rediscovery 89.6% | Not checked for this page |
| Claude Opus 5.5 | FrontierSWE v2 62.3%; AutomationBench 42.5%; Terminal-Bench-Science 0.1 58.7%; OSWorld 2.1 (offline, partial) 87.2%; Humanity's Last Exam (with tools) 67.7%; CVE Rediscovery 81.3% (without safeguards) | Not checked for this page |
The Opus 5.5 announcement (22 September 2026) uses different suites and compares Opus 5.5 with Fable 5.1 and Opus 5, for example Terminal-Bench 4.0 66.4% (Opus 5.5), 55.8% (Fable 5.1) and 52.3% (Opus 5). Figures from different announcements are not directly comparable. Run your own evaluation before choosing.
Deprecations and retirements
As of 2026-11-15. Source: Anthropic model deprecations page. Anthropic gives at least 60 days' notice before retiring publicly released models; "not sooner than" dates apply to Anthropic-operated platforms, and Amazon Bedrock and Google-managed Google Cloud set their own schedules.
| Model | API ID | Notice or deprecation date | Retirement or shutdown date | Stated replacement |
|---|---|---|---|---|
| Claude Fable 5.2 | claude-fable-5-2 | Active | Not sooner than 2027-11-09 | None stated |
| Claude Fable 5.1 | claude-fable-5-1 | Active | Not sooner than 2027-09-01 | None stated |
| Claude Fable 5 | claude-fable-5 | Active | Not sooner than 2027-06-09 | None stated |
| Claude Mythos 5.2 | claude-mythos-5-2 | Active | Not sooner than 2027-11-09 | None stated |
| Claude Mythos 5.1 | claude-mythos-5-1 | Active | Not sooner than 2027-09-01 | None stated |
| Claude Mythos 5 | claude-mythos-5 | Active | Not sooner than 2027-06-09 | None stated |
| Claude Mythos Preview | claude-mythos-preview | 2026-06-09 | To be announced | None stated |
| Claude Opus 5.5 | claude-opus-5-5 | Active | Not sooner than 2027-09-22 | None stated |
| Claude Opus 5 | claude-opus-5 | Active | Not sooner than 2027-07-24 | None stated |
| Claude Opus 4.8 | claude-opus-4-8 | Active (fast mode removal 2026-11-30) | Not sooner than 2027-05-28 | Opus 5.5 named for fast mode migration |
For lifecycle planning, see the AI Model Retirement Guide and the LLM Deprecation and Retirement Tracker.
Data handling and security
Anthropic's zero data retention page lists six Covered Models: Fable 5.2, Mythos 5.2, Fable 5.1, Mythos 5.1, Fable 5 and Mythos 5. They "require 30-day data retention", so zero data retention is not available for them unless Anthropic expressly authorises it. The Opus 5.5 announcement states that zero data retention is available, as with previous Opus models; the retention page directs readers to their contract terms for current eligibility. Under any arrangement, flagged chats or sessions may be retained for up to 2 years.
Mythos access is tied to organisational verification, so the credential that calls claude-mythos-5-2 carries more weight than a general key: a leaked key from a verified organisation exposes restricted capability as well as spend. Store Claude API keys in a secrets manager, scope workspaces per use case and alert on unusual model IDs. See the Anthropic API Keys Security Guide, the LLMjacking Guide and the Claude for Enterprise Security Guide.
How to choose a model: factual criteria
| Criterion | What the documentation shows |
|---|---|
| Access | Fable and Opus are open to all customers; every numbered Mythos model requires verification |
| List price per token | $4/$20 Opus 5.5; $5/$25 Opus 5 and 4.8; $8/$40 Fable and Mythos 5.2; $10/$50 Fable and Mythos 5 and 5.1 |
| Retention | 30-day retention required for Fable and Mythos 5.x; ZDR stated as available for Opus 5.5 |
| Default effort | High on Fable 5.2 and Mythos 5.2; medium on Opus 5.5 |
| Cyber and biology requests | Fable 5.2 routes blocked cyber requests to Opus 4.8 and biology requests to Opus 5; Opus 5.5 routes most cyber tasks to Opus 4.8 |
| Lifecycle runway | Fable 5.2 and Mythos 5.2 have the latest "not sooner than" date (2027-11-09) |
Practitioner checklist
- Pin production code to an explicit API ID (claude-fable-5-2, claude-mythos-5-2 or claude-opus-5-5) and record it in the AI inventory.
- Confirm 30-day retention is acceptable before routing data to any Fable or Mythos 5.x model.
- Test migrations for 400 errors from forced tool use, sampling parameters, disabled thinking and the Fable 5.2 prefix check.
- Expect some Fable 5.2 and Opus 5.5 security requests to be answered by Opus 4.8 or Opus 5, and log the responding model.
- Move any Opus 4.8 fast mode traffic before 30 November 2026.
- Restrict who can call Mythos model IDs inside a verified organisation and review that list quarterly.
- Re-measure token counts and thinking output on real prompts before comparing Fable and Opus cost.
- Check partner cloud catalogues separately for their own model IDs and retirement dates.
Standards and references
- Anthropic: Models overview (accessed 2026-11-15)
- Anthropic: Pricing (accessed 2026-11-15)
- Anthropic: Model deprecations (accessed 2026-11-15)
- Anthropic: Claude Platform release notes (accessed 2026-11-15)
- Anthropic: Zero data retention (accessed 2026-11-15)
- Anthropic: Claude Fable 5.2 model page (accessed 2026-11-15)
- Anthropic: Claude Mythos 5.2 model page (accessed 2026-11-15)
- Anthropic: Claude Opus 5.5 model page (accessed 2026-11-15)
- Anthropic: Claude Fable 5.2 and Claude Mythos 5.2 announcement (accessed 2026-11-15)
- Anthropic: Introducing Claude Opus 5.5 (accessed 2026-11-15)
Related NHI Mgmt Group resources: Anthropic Models, Claude Opus 5.5 vs Opus 5 vs Opus 4.8, Claude Mythos Versions Compared, Frontier AI Models Compared, LLM Pricing Tracker.