Breaking

Anthropic launches Claude Fable 5.1 and invitation-only Claude Mythos 5.1

September 1, 2026 at 14:11 EDT

Foundation Models
AI Agents
Software Dev & Coding

Anthropic has released Claude Fable 5.1 and Claude Mythos 5.1, positioning the pair as its most capable models for coding and knowledge work, with stronger scores on long-horizon agent tasks and cheaper prompt-cache reads. The company says the two products share the same underlying model; Fable 5.1 is generally available, while Mythos 5.1 is limited to trusted-access programs for cybersecurity and life-sciences work.

September 1, 2026 · Anthropic

Fable 5.1 raises the agent bar

Claude Fable 5.1 is generally available; Claude Mythos 5.1 is invitation-only. Same weights, cheaper cache reads, and a large jump on scientific terminal tasks.

52.6%

Terminal-Bench-Science 0.1

55.8%

Terminal-Bench 4.0

$0.25

Cache read / MTok

~25%

Typical billed savings

Terminal-Bench-Science 0.1 — resolution rate

22.4%

GPT-5.6 Sol

24.7%

Fable 5

29.0%

Opus 5

52.6%

Fable 5.1

Fable 5.1 more than doubles Fable 5 on this science-agent suite (Anthropic eval; standard error ±3.5–4.5 pts).

What testers liked

Long unattended runs, cheaper per-task cost after cache-read cuts, and stronger coding/debug traces than Fable 5.

Limits to watch

Forced tool use is gone; thinking blocks bind to this model; science scores still leave most tasks unsolved.

Same list price, cheaper cache path

Input/output stay $10 / $50 per million tokens. Cache reads drop 75% to $0.25/MTok. Anthropic estimates ~25% lower typical bills and up to ~45% on highly agentic sessions.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.