Claude Fable 5: The Ultimate Guide to Anthropic's Most Powerful AI
The first Mythos-class model available to the public. Discover benchmarks, API pricing, autonomous coding capabilities, and how Fable 5 compares to GPT-5.5 and Gemini 3.1 Pro.
What is Claude Fable 5?
On June 9, 2026, Anthropic made history by releasing the first Mythos-class AI model to the general public. Here is everything you need to know.
Claude Fable 5 is not merely an incremental update in Anthropic's Opus lineage — it sits entirely above it. It represents Anthropic's most capable widely released model to date, specifically engineered for the most demanding reasoning tasks and long-horizon agentic work. The core philosophy is simple yet profound: the longer and more complex the task, the larger Fable 5's lead over any other model on the market.
The model is designed to handle multi-day autonomous sessions, massive codebase migrations, and intricate knowledge synthesis that would cause previous generation models to hallucinate or lose context. It supports a 1,000,000 token context window and can generate up to 128,000 output tokens per request.
Fable 5 introduces a fascinating dual-model paradigm. The underlying neural network weights are identical to those of Claude Mythos 5. The critical distinction lies entirely in the safety architecture — Mythos 5 remains restricted to vetted partners, while Fable 5 is the safe-for-everyone twin with dynamic safety classifiers.
Key Insight: Anthropic named the model after the Latin fabula ("that which is told"), a cousin of the Greek mythos. The safeguards are the only thing separating the two models.
Anthropic Mythos-Class · June 9, 2026
Claude Fable 5
The public-safe Mythos-class model. Includes dynamic safety classifiers that route high-risk queries (cybersecurity, biology) to Opus 4.8. Available on Claude API, Amazon Bedrock, Google Vertex AI, and Microsoft Foundry. API ID: claude-fable-5
Claude Mythos 5
The same underlying model as Fable 5, but with certain safety classifiers disabled. Available only to vetted partners through Project Glasswing — primarily government cyber-defenders and approved biology researchers. Not commercially available.
Benchmarks: Fable 5 vs GPT-5.5 vs Gemini 3.1 Pro
Claude Fable 5 dominates on nearly every major benchmark, with the most pronounced leads in complex, long-horizon software engineering and knowledge work.
| Benchmark | Category | Claude Fable 5 | Claude Opus 4.8 | GPT-5.5 | Gemini 3.1 Pro |
|---|---|---|---|---|---|
| SWE-Bench Pro | Agentic Coding | 80.3% | 69.2% | 58.6% | 54.2% |
| SWE-Bench Verified | Coding | 95% | 88.6% | — | 80.6% |
| FrontierCode Diamond | Production Coding | 29.3% | 13.4% | 5.7% | — |
| GDPval-AA (ELO) | Knowledge Work | 1932 | 1890 | 1769 | 1314 |
| GDP.pdf Vision | Document Vision | 29.8% | 22.5% | 24.9% | 16.7% |
| Blueprint-Bench 2 | Spatial Reasoning | 38.6% | 14.5% | 36.2% | 26.5% |
| AutomationBench | Tool Use | 17.4% | 15.5% | 12.9% | 9.6% |
| OSWorld-Verified | Computer Use | 85% | 83.4% | 78.7% | 76.2% |
| Legal Agent Bench | Legal Reasoning | 13.3% | 10.4% | 2.1% | 0% |
SWE-Bench Pro: Agentic Coding
Multi-Benchmark Analysis
Important Note: Some published benchmark scores (particularly for ExploitBench and HealthBench Professional) reflect the performance of the restricted Mythos 5 model. Because Fable 5's safety classifiers actively block offensive cybersecurity and biology queries, its real-world performance on those specific tasks is intentionally closer to Claude Opus 4.8. For all standard enterprise use cases — coding, data analysis, legal reasoning, and automation — Fable 5 delivers the full Mythos-class performance shown above.
The Era of Autonomous Software Engineering
Claude Fable 5 is purpose-built for long-horizon, autonomous coding tasks that previous models simply could not sustain.
Large-Scale Migrations
Fable 5 can execute codebase-wide migrations across millions of lines of code. Stripe confirmed it completed a 50M-line Ruby migration in a single day — a task estimated at over two months for a full engineering team.
Multi-Day Autonomous Sessions
When deployed in Claude Code or agent harnesses, Fable 5 can work unattended for days. It plans across stages, delegates to sub-agents, writes its own tests, and verifies outputs against original design goals.
Self-Verification & Correction
At high effort settings, Fable 5 reflects on and validates its own work. It can catch complex design holes, run its own test suites, and use vision to compare rendered outputs against design specifications.
Vision-Powered Development
Fable 5 can rebuild a web application's source code from screenshots alone. It uses vision to evaluate rendered UI against design goals — a capability that eliminates entire categories of manual QA work.
Enterprise Knowledge Work
From dense financial documents to complex legal contracts, Fable 5 excels at multi-stage knowledge work. IMC reported it aced trading-analysis evaluations across factual lookup, root-cause analysis, and expected-value analysis.
Scientific Research
Fable 5 is Anthropic's first model to consistently produce novel, compelling scientific hypotheses. In blinded comparisons, scientists preferred Mythos-class hypotheses ~80% of the time over Opus-class outputs.
“Fable 5 compresses months of engineering into days. In our 50-million-line Ruby codebase, it did in a day what would've taken us more than two months by hand.”
How to Use Claude Fable 5 for Coding via API
Getting started with Claude Fable 5 for your coding projects is straightforward. Use the model ID claude-fable-5 in your API calls. The model supports the same prompting techniques as other Claude models, with some optimizations for long-context tasks.
import anthropic
client = anthropic.Anthropic()
# Initialize a long-horizon coding task with Claude Fable 5
message = client.messages.create(
model="claude-fable-5",
max_tokens=16000,
messages=[
{
"role": "user",
"content": "Analyze this 50k-line codebase and plan a migration strategy."
}
]
)
print(message.content)Claude Fable 5 API Pricing & Access Guide
Fable 5 is a premium model priced at twice the cost of Opus 4.8. Understanding the token economics is essential for cost-effective deployment.
Claude Haiku
$4 / M output tokens
- ✓Fastest response times
- ✓Best for high-volume tasks
- ✓Simple Q&A and classification
- ✓Cost-optimized workflows
Claude Opus 4.8
$25 / M output tokens
- ✓Strong general performance
- ✓Fable 5's fallback model
- ✓Best price-to-capability ratio
- ✓Ideal for most enterprise tasks
Claude Fable 5
$50 / M output tokens
- ✓Mythos-class intelligence
- ✓1M token context window
- ✓90% prompt caching discount
- ✓Multi-day autonomous sessions
- ✓SWE-Bench Pro: 80.3%
- ✓Available on all major clouds
Prompt Caching Discount: Fable 5 offers a 90% discount on cached input tokens. For agentic workflows that repeatedly send the same large system prompt or codebase context, this can reduce your effective input cost from $10/M to just $1/M on cache hits. Cache writes cost $12.50/M (5-min) or $20/M (1-hour). Designing your application to maximize cache hits is the single most impactful cost optimization for Fable 5 deployments.
API Pricing Comparison
Understanding the Safeguard Fallback System
Fable 5's dynamic safety classifiers are a unique architectural feature every developer must understand before deploying in production.
User Sends Request
Your application sends a prompt to the claude-fable-5 model via the API.
Safety Classifier Runs
Fable 5's real-time classifiers analyze the prompt for cybersecurity, biology/chemistry, or distillation-related content.
Safe: Fable 5 Responds
In 95%+ of sessions, no flag is triggered. Fable 5 responds with full Mythos-class capability.
Flagged: Fallback to Opus 4.8
If flagged, the request is automatically routed to Opus 4.8. You are NOT charged Fable 5 rates for this response.
For developers, the safeguard system has important billing implications. When a fallback occurs, the Messages API returns a successful HTTP 200 response, but the stop_reason field will indicate "refusal". You are only billed for the Opus 4.8 usage in these instances, not Fable 5 rates.
Additionally, all Mythos-class traffic (including Fable 5) requires a mandatory 30-day data retention period for Anthropic's safety monitoring. This is a non-negotiable policy and means Fable 5 is not available under zero data retention agreements.
What Industry Leaders Are Saying
Companies with early access to Claude Fable 5 have shared their real-world results.
“Claude Fable 5 is the state of the art model on CursorBench. It's opened up a class of long-horizon problems that were out of reach for earlier models.”
“Claude Fable 5 is the first to break 90% on our core analytics benchmark of complex, long-running analytical tasks — a 10-point jump over Opus.”
“Claude Fable 5 came out #1 on our evals, winning head-to-head against every model we tested. Significantly stronger on UI design and game coding.”
“Claude Fable 5 is a real step forward for developers. It took on complex, long-horizon coding tasks with a level of autonomy that exceeded previous benchmarks.”
“Claude Fable 5 is the strongest finance-first model we've tested, both on general finance and reasoning. It's a notable step up.”
“At the highest effort, Claude Fable 5 reflects on and validates its own work. For us, that's what makes highly autonomous operations possible.”
Frequently Asked Questions
Everything you need to know about Claude Fable 5, from technical details to pricing and access.
Claude Fable 5 is Anthropic's most capable publicly available AI model, released on June 9, 2026. It is the first model from Anthropic's "Mythos-class" tier to be made generally available to developers and enterprises. The model is designed for long-horizon agentic work, complex software engineering, and demanding knowledge tasks that previous models could not sustain over extended periods.
The name "Fable" comes from the Latin fabula, meaning "that which is told" — a deliberate linguistic cousin of the Greek mythos, reflecting the model's shared intelligence with the restricted Mythos 5 model.
Fable 5 and Mythos 5 share the exact same underlying neural network weights and core intelligence. The difference is entirely based on access control and safety guardrails. Fable 5 is the publicly available version with dynamic safety classifiers that block high-risk queries (cybersecurity exploits, biological hazards, model distillation) and route them to Claude Opus 4.8.
Mythos 5 is a restricted model with some of those safety classifiers disabled. It is only available to vetted partners through Project Glasswing — primarily US government cyber-defenders and approved biology researchers. For all standard enterprise and developer tasks, Fable 5 performs effectively identically to Mythos 5.
Claude Fable 5 is priced at $10 per million input tokens and $50 per million output tokens. This is exactly double the cost of Claude Opus 4.8 ($5/$25). A 90% prompt caching discount applies to cached input tokens, reducing the effective input cost to $1/M on cache hits.
For US-only inference (data residency requirements), pricing is available at a 1.1x multiplier. On Claude.ai subscription plans, Fable 5 counts as 2x usage. Anthropic also offered a free promotional window for Pro, Max, Team, and Enterprise subscribers from June 9–22, 2026.
Based on Anthropic's published benchmarks and extensive third-party testing, Claude Fable 5 significantly outperforms GPT-5.5 on complex reasoning and software engineering tasks. On SWE-Bench Pro, Fable 5 scored 80.3% compared to GPT-5.5's 58.6% — a gap of over 21 percentage points. On FrontierCode Diamond, Fable 5 scored 29.3% versus GPT-5.5's 5.7%.
GPT-5.5 remains competitive in specific niches: it performs well on Terminal-Bench 2.1 via its Codex CLI harness (83.4%) and is competitive on spatial reasoning (Blueprint-Bench 2: 36.2% vs Fable 5's 38.6%). GPT-5.5 also offers a lower price point ($5/$30 vs $10/$50). For long-horizon autonomous coding and complex knowledge work, however, Fable 5 is the clear leader.
Claude Fable 5 supports a 1,000,000 token (1M token) context window by default. It also supports generating up to 128,000 output tokens per single request. This massive context window enables Fable 5 to process entire codebases, lengthy legal documents, or extensive research papers in a single session without losing context.
This is the intended behavior of Fable 5's safeguard system. If your prompt contained keywords or concepts related to offensive cybersecurity (e.g., building exploits or malware), creating biological hazards, or model distillation, Fable 5's safety classifiers intercepted the request and automatically routed it to Claude Opus 4.8.
Importantly, you are not charged Fable 5 rates when this fallback occurs — you only pay for the Opus 4.8 usage. The API will return a successful HTTP 200 response with stop_reason: "refusal" in the metadata to indicate the fallback occurred. Anthropic reports this happens in fewer than 5% of average sessions.
You should reach for Claude Fable 5 when dealing with tasks that require sustained autonomy, complex multi-step reasoning, or massive code refactoring. Ideal use cases include: large-scale codebase migrations, multi-day autonomous coding sessions, complex financial document analysis, multi-stage research synthesis, and any task where the AI needs to plan, execute, and self-verify over an extended period.
You should stick with Claude Opus 4.8 for routine, high-volume, or latency-sensitive work: classification, summarization, drafting, interactive chat, and most day-to-day agentic tasks. At half the price, Opus 4.8 is the sensible default for the majority of enterprise workloads. The optimal strategy for most organizations is intelligent routing — using Fable 5 for the hardest 20% of tasks and Opus 4.8 for the rest.
Claude Fable 5 is available across multiple platforms as of June 9, 2026. For developers, it is accessible via the Claude API directly (model string: claude-fable-5), as well as through Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. For enterprise customers, it is available on the consumption-based Enterprise plan.
For individual users, Fable 5 is accessible through the Claude.ai web interface and mobile apps on paid plans (Pro, Max, Team, and Enterprise). Note that Fable 5 counts as 2x usage on subscription plans due to its higher compute requirements.