DataLearner logo

Claude Fable 5vsClaude Mythos Preview

Across 4 shared benchmarks, Claude Fable 5 leads overall: Claude Fable 5 wins 3, Claude Mythos Preview wins 1, with 0 ties and an average score difference of +0.83.

Anthropic
Claude Fable 5

Anthropic · 2026-06-09 · Reasoning model

Anthropic
Claude Mythos Preview

Anthropic · 2026-04-07 · Chat model

Claude Fable 53 wins(75%)(25%)1 winClaude Mythos Preview

Benchmark scores

Grouped by capability, sorted by largest gap within each. 4 shared benchmarks.

Coding and Software Engineer

Claude Fable 5 2/2
BenchmarkClaude Fable 5Claude Mythos PreviewDiff
SWE-Bench Pro - Public80.301 / 57Deep Thinking (With Tools)77.803 / 57Extended (with tools)+2.50
SWE-bench Verified952 / 114Deep Thinking (With Tools)93.904 / 114Extended (with tools)+1.10

AI Agent - Tool Usage

Claude Fable 5 1/1
BenchmarkClaude Fable 5Claude Mythos PreviewDiff
OSWorld-Verified851 / 26Thinking High (With Tools)79.608 / 26Extended (with tools)+5.40

General Knowledge

Claude Mythos Preview 1/1
BenchmarkClaude Fable 5Claude Mythos PreviewDiff
HLE595 / 181Deep Thinking (No Tools)64.701 / 181Extended (with tools)-5.70

Specs

FieldClaude Fable 5Claude Mythos Preview
PublisherAnthropicAnthropic
Release date2026-06-092026-04-07
Model typeReasoning modelChat model
ArchitectureDenseDense
ParametersNot availableNot available
Context length1MNot available
Max output128K8K

API pricing

Prices use DataLearner records when available; missing fields are not inferred.

ItemClaude Fable 5Claude Mythos Preview
Text input$10 / 1M tokens$25 / 1M tokens
Text output$50 / 1M tokens$125 / 1M tokens
Cache read$1 / 1M tokensNot public
Cache write$12.5 / 1M tokensNot public

Summary

  • Claude Fable 5leads in:Coding and Software Engineer (2/2), AI Agent - Tool Usage (1/1)
  • Claude Mythos Previewleads in:General Knowledge (1/1)

On average across the 4 shared benchmarks, Claude Fable 5 scores 0.83 higher.

Largest single-benchmark gap: HLE — Claude Fable 5 59 vs Claude Mythos Preview 64.70 (-5.70).

Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.