Claude Mythos 5.1
Anthropic's frontier-tier model designed for advanced cybersecurity and critical infrastructure work
Context window
1M
Input / 1M tokens
$10.00
Output / 1M tokens
$50.00
Provider
Anthropic
Data verified 2026-09-23
Claude Mythos 5.1 is Anthropic's most capable model tier, sitting above Claude Opus. It excels at finding software vulnerabilities and advanced reasoning tasks. Available only in limited release through Project Glasswing, it shares identical capabilities and pricing with Claude Fable 5.1 but without safety classifiers.
Capability index
Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.
How to access it
Limited access through Project Glasswing, restricted to approved cybersecurity and critical infrastructure organizations. Contact Anthropic, AWS, or Google Cloud account team for access.
Strengths
- ✓Exceptional cybersecurity vulnerability detection (93.9% SWE-bench, 97.6% USAMO)
- ✓Long context reasoning (1M token window)
- ✓Advanced code generation and analysis
Best for developers who...
When to choose it (and when not to)
Reach for Claude Mythos 5.1 when...
- →When you have approved access through Project Glasswing
- →For cybersecurity research and infrastructure hardening
- →When maximum capability is needed for frontier tasks
Look elsewhere if...
- ✕If you do not have Project Glasswing approval
- ✕For general-purpose use (Claude Fable 5.1 offers same capabilities with safeguards)
- ✕If you need general availability and public pricing
How to use it
- ›Use structured reasoning prompts for vulnerability analysis
- ›Leverage full 1M context window for comprehensive codebase review
Quickstart
Pythonimport anthropic
client = anthropic.Anthropic()
message = client.messages.create(
model="claude-mythos-5-1",
max_tokens=1024,
messages=[{"role": "user", "content": "Analyze this code for vulnerabilities..."}]
)
print(message.content)Requires Project Glasswing access; standard Claude API credentials will not work
API model id: claude-mythos-5-1
Benchmarks
| Benchmark | Score | Notes |
|---|---|---|
| SWE-bench Verified | 93.9% | Software engineering task completion |
| USAMO 2026 | 97.6% | Mathematical olympiad problems |
| Terminal-Bench 2.0 | 82.0% | Command-line reasoning tasks |
Source: Anthropic official announcement
Compare Claude Mythos 5.1
Compare Claude Mythos 5.1 with any other model
Build a comparison →All model comparisons →