NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/Claude Mythos Preview
Anthropic logo

Claude Mythos Preview

by Anthropic · Released Apr 2026

1M ContextPreview
99.8
avg score
Rank #2
Compare
Top 1% of all models
Context
1.0M tokens (~500 books)
Input $/1M
n/a
Output $/1M
n/a
Type
text
License
Proprietary
Benchmarks
14 tested
Data refreshed daily
About

Anthropic's most capable model · Claude Mythos Preview. Tops SWE-bench Verified (93.9%), GPQA Diamond (94.5%), USAMO (97.6%), and HLE with tools (64.7%). Adaptive thinking at max effort, context up to 1M tokens.

Tested on 14 benchmarks · BenchGecko score 99.8. Top scores: USAMO (97.6%), GPQA diamond (94.5%), SWE-Bench verified (93.9%).

Capabilities
coding
80.0
#7 globally
reasoning
81.0
#15 globally
math
97.6
#1 globally
knowledge
81.3
#5 globally
agentic
79.6
#1 globally
Benchmark Scores
Compare All
Tested on 14 benchmarks · Ranked across 5 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
codingCompare coding →
SWE-Bench verified

Real-world software engineering tasks from GitHub issues. Models must diagnose bugs and write patches that pass test suites. Human-verified subset of SWE-bench.

93.9·
SWE-bench Multilingual

SWE-bench extended to non-Python languages. Tests coding ability across Java, JS, Go, Rust, and more.

87.3·
Terminal Bench

Complex terminal-based engineering tasks. Models must use command-line tools, navigate filesystems, and debug systems through shell interaction.

82.0·
reasoningCompare reasoning →
CharXiv Reasoning (with tools)

Chart reasoning with tool use. Models can use code execution to analyze scientific figures.

93.2·
CharXiv Reasoning

Chart and figure reasoning from arXiv papers. Tests ability to interpret scientific visualizations.

86.1·
GraphWalks BFS 256K-1M

Graph traversal benchmark at 256K context. Tests ability to follow breadth-first search paths in large graph structures.

80.0·
mathCompare math →
USAMO

USA Mathematical Olympiad problems. Among the hardest math competitions, requiring elegant proofs and deep mathematical insight.

97.6·
Quick compare:OpenAI logovs GPT-5.5 ProAnthropic logovs Claude Opus 5.5OpenAI logovs GPT-6 Astra
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Similar Models
OpenAI logo
GPT-5.5 Pro
OpenAI
99.9$30.00/1M
Anthropic logo
Claude Opus 5.5
Anthropic
96.8$4.00/1M
OpenAI logo
GPT-6 Astra
OpenAI
96.3$10.00/1M
Links
Info
AnthropicPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@Anthropic
BenchGecko API
claude-mythos-preview
Specifications
  • Typetext
  • Context1.0M tokens (~500 books)
  • ReleasedApr 2026
  • LicenseProprietary
  • Statuspreview
Available On
Anthropic logoAnthropicn/a
Categories
codingreasoningmathknowledgeagentic
Learn More
context-windowtransformertokens
Share & Export
Tweet
Related Models
OpenAI logoGPT-5.5 Pro
99.9
Anthropic logoClaude Opus 5.5
96.8
OpenAI logoGPT-6 Astra
96.3
DeepSeek logoDeepSeek V3.2 Speciale
95.2
Claude Mythos Preview is a proprietary text AI model by Anthropic, released in April 2026. It has an average benchmark score of 99.8. Context window: 1M tokens.

Related Models

GPT-5.5 Pro · OpenAIClaude Opus 5.5 · AnthropicGPT-6 Astra · OpenAIDeepSeek V3.2 Speciale · DeepSeekStep 3.5 Flash · stepfun

Benchmarks

USAMOGPQA diamondSWE-Bench verifiedCharXiv Reasoning (with tools)MMMLU

Related Pages

Anthropic · ProviderAnthropic · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts

  • Claude Mythos Preview by Anthropic. BenchGecko score 99.8, rank 2 of 312 scored models (normalized average of public benchmark scores).
  • List price n/a input · n/a output per 1M tokens.

How to cite

Claude Mythos Preview · benchmarks, pricing and providers. BenchGecko. https://benchgecko.ai/model/claude-mythos-preview

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP