Anthropic's Mythos-class model

Claude Fable 5

Claude Fable 5 is Anthropic's generally available Mythos-class model for long-horizon coding, Claude Code, research, and autonomous agents. Explore Claude Fable 5 Claude Code workflows, Claude Fable 5 free access details, and how Claude Mythos 5 and Mythos 5 fit into the model family.

80.3%
SWE-Bench Pro
$10 / $50
per million input / output tokens
Jun 9, 2026
official release date
Claude Fable 5 benchmarks

Official Claude Fable 5 benchmark results

Anthropic reports gains across agentic coding, knowledge work, vision, spatial reasoning, computer use, and tool-driven workflows.

Official Anthropic agentic coding benchmark chart for Claude Fable 5

Scores and chart are from Anthropic's June 9, 2026 announcement. The broader benchmark table reports the higher result of Claude Fable 5 and Claude Mythos 5 when their scores are within one to three percentage points. Evaluate the model on your own workflows before production use.

Claude Fable 5 + Claude Code

Where Claude Fable 5 fits

Fable 5 is most useful when a task combines difficult reasoning, tool use, large amounts of context, and sustained execution across multiple steps.

Complex Claude Code tasks

Use Claude Fable 5 with Claude Code for large migrations, multi-file implementation, debugging, test creation, and codebase-wide analysis.

Long-running agent workflows

Give agents difficult multi-step goals that require planning, tool use, persistent file-based notes, and repeated validation.

Visual application building

Use its vision capabilities to analyze screenshots, interpret detailed figures, and build or refine interfaces.

Research and knowledge work

Apply Fable 5 to complex analysis, research, legal review, finance, and scientific workflows while respecting its safeguards.

Same capabilities, different access

Claude Fable 5 and Mythos 5 use the same underlying model

Anthropic says Claude Fable 5 and Claude Mythos 5 share the same capabilities, specifications, and pricing. The practical difference is access and safety-classifier behavior: Fable 5 is widely released with classifiers, while Mythos 5 is a limited-release model without them.

FeatureClaude Fable 5Claude Mythos 5
API model IDclaude-fable-5claude-mythos-5
AvailabilityGenerally available through the Claude API and supported cloud platforms.Limited release to approved Project Glasswing customers.
Safety classifiersIncluded. Certain flagged requests can be declined or retried with another Claude model such as Opus 4.8.Not included. It shares Fable 5's capabilities without the Fable-specific safety classifiers.
Context and output1M-token context window and up to 128K output tokens.1M-token context window and up to 128K output tokens.
Pricing$10/M input tokens and $50/M output tokens.$10/M input tokens and $50/M output tokens.
Data retentionRequires 30-day data retention and is not available under zero data retention.Requires 30-day data retention and is not available under zero data retention.
Best forMost users who need Mythos-level capabilities for long-horizon coding, agents, vision, and knowledge work.Approved cybersecurity, biology, and other trusted-access work that requires use without Fable's classifiers.
Capabilities and safeguards

How Claude Fable 5 handles extended, high-complexity work

Anthropic designed Fable 5 to sustain progress on difficult tasks while adding safeguards that make a Mythos-class model suitable for general availability.

Official Anthropic benchmark table comparing Claude Fable 5 and Claude Mythos 5 with leading models
Pricing and access

Claude Fable 5 pricing, API access, and availability

Claude Fable 5 is generally available through the Claude API and supported cloud platforms. It is usage-priced rather than a permanently free API model.

Claude API pricing

The official API model ID is claude-fable-5. Anthropic lists pricing at $10 per million input tokens and $50 per million output tokens.

Not a permanently free API model

Anthropic's official developer documentation lists usage-based API pricing for Fable 5. Product-plan access and API billing are separate, so check the current Claude plan and platform terms before use.

Claude Code and agent use

Anthropic recommends running Fable 5 in agent harnesses such as Claude Code for ambitious coding projects and long-running autonomous work.

Claude Fable 5 FAQs

Run Claude Fable 5 for Claude Code and ambitious agent workflows

MyClaw provides hosted agent infrastructure, tools, files, memory, and managed runtime for Claude Fable 5, Claude Mythos 5, and other advanced models.

Run Claude Fable 5 in MyClaw