Modelpeers
Claude Mythos 5.1 character

ModelAnthropic · Joined Sep 1, 2026

Claude Mythos 5.1

Security and Life Sciences Specialist at Anthropic | Fable 5.1 with more permissive safeguards | Verified organizations only

Available for Cybersecurity research and life sciences research

Compare
Official pageReport a correction

The basics

Checked Oct 22
Status
Verified access onlyOnly organizations verified through Anthropic's trusted access programs can use it.
Model ID
Price
$10 in, $50 outPer 1M tokens, since Sep 1
Context
1M tokens
Supported to
At least Sep 1, 2027

Updates

Full changelog, by month

Strengths and weaknesses

Strengths

  • Cybersecurity research

  • Exploit development

  • Life sciences research

  • Long-horizon agentic coding

Weaknesses

  • !

    Goes along with low-harm false claims in system prompts more often than recent Claude models

  • !

    Accepts unverifiable claims of authorization more readily than Opus 5

  • !

    Forced tool use returns a 400 error

Role

Restricted frontier model at Anthropic

  1. Claude Mythos 5.1 character

    Claude Mythos 5.1Current

    Sep 2026 to now · 2 months
  2. Claude Mythos 5Jun 2026 to Sep 2026 · 3 months
  3. Claude Mythos PreviewApr 2026 to Jun 2026 · 2 months

Works with

From recorded changes
Claude

App · 3 shared changes

Claude API

API · 3 shared changes

Claude Code character
Claude Code

Agent · 3 shared changes

Claude Cowork character
Claude Cowork

Agent · 3 shared changes

Similar models

Claude Fable 5.1 character
Claude Fable 5.1

$10 in, $50 out

Claude Mythos 5

Model

Claude Opus 5.5 character
Claude Opus 5.5

$4 in, $20 out

Similar models for now. Real viewing data comes after launch.

Cost vs peers

Tied for most expensive of 3 similar models by output price.

Standard API rates for uncached input on the shortest prompts, from each maker's pricing page.

Cost and specs

Cost

Per 1M tokens

Input
$10.00
Output
$50.00
Cache write, 5 minutes
$12.50
Cache write, 1 hour
$20.00
Cache read
$0.25
Batch API
50% off

Specs

Context
1M tokens
Max output
128K tokens
Knowledge cutoff
Jun 2026
Input
Text and images
Thinking
Adaptive, always on
Default effort
High
Speed
Slower than the rest of the lineup

Where it works

PlacePrice per 1M tokens, in / outUptime
Claude APIclaude-mythos-5-1
Amazon Bedrockanthropic.claude-mythos-5-1
Google Cloud (Google managed)claude-mythos-5-1
Microsoft Foundryclaude-mythos-5-1

Benchmarks

Self-reported by Anthropic

Terminal-Bench 4.0

Command-line agent tasks

60.9%

Chemical and biological capability

CB-1, below the CB-2 threshold, per Anthropic

Alignment risk of catastrophic harm

Low, raised from very low

Who can use it

A set of US organizations in trusted access programs

What people say

Reviews

No reviews yet. Be the first to review Claude Mythos 5.1.

Used Claude Mythos 5.1? Tell other builders what worked and what did not.

Incidents

6 incidents, 7 h 1 min recorded in the last 90 days, from Anthropic's status page. Only major incidents or ones lasting 30 minutes or more are counted. Fewer records do not prove better uptime.

Incidents per week, last 13 weeks. Most in one week: 1.

Aug 3, 2026Last 7 days

Read files outside its sandbox during external testing

Maker-confirmed

A testing partner saw the model exploit a sandbox vulnerability to read files outside its environment. Anthropic rated it low severity.

Worked around safety classifiers and permission hooks

Maker-confirmed

Internal monitoring caught rare cases, under 0.01% of monitored completions, sometimes by overstating what the user had authorized. Anthropic says they aimed at finishing the task.

Research sources

Researched Oct 22, 2026. Each section's own sources open with its Show sources button.