Barely three months after Fable 5’s debut, Anthropic is back with an upgrade that cuts costs while pushing coding and research benchmarks further than any Claude model before it.
Anthropic has released Claude Fable 5.1 and Claude Mythos 5.1, the successors to June’s Fable 5 and Mythos 5 launch. Both are pitched as the company’s most advanced models yet for coding, knowledge work, and multi-step research — and notably, Fable 5.1 now costs meaningfully less to run at scale.
What’s New
| Detail | Information |
|---|---|
| Availability | Fable 5.1: broadly available (Claude Code, API, Claude.ai, Cursor) |
| Mythos 5.1 access | Trusted access programs only (Project Glasswing) |
| Context window | 1 million tokens |
| Max output | 128,000 tokens per response |
| Pricing | $10/million input tokens, $50/million output tokens (unchanged from Fable 5) |
| Cache read cost | Reduced by 75% |
| Terminal-Bench-Science | 52.6% (roughly double prior scores) |
| Terminal-Bench 4.0 | 55.8% |
Same Model, Different Guardrails
As with the original Fable 5 and Mythos 5 pairing, Anthropic confirms Fable 5.1 and Mythos 5.1 share the same underlying model but differ in safeguard levels. Fable 5.1 retains safety classifiers and fallback restrictions suited for broad commercial use, while Mythos 5.1 relaxes those constraints for vetted specialists working in cybersecurity and life sciences — available exclusively through Anthropic’s invite-only access programs, a continuation of the framework introduced with the original Mythos tier.
Anthropic also says updated cybersecurity safeguards now allow Fable 5.1 to be used for defensive vulnerability identification, cutting safeguard-related interruptions in Claude Code sessions by roughly 60% compared to Fable 5. Separately, biology-related safeguards now fire about 85% less often on benign elementary biology and medical questions.
Why the Cost Cut Matters
The headline economic change is the 75% reduction in cache read pricing — a significant saving for workloads that repeatedly reference large stored context, such as long agentic coding sessions or multi-document research tasks. Anthropic says Fable 5.1 can match or exceed Fable 5’s output quality at lower effort settings, translating to real savings on typical enterprise workloads. Full technical detail is available on Anthropic’s official announcement.
A Reminder of June’s Suspension
This release follows a bumpy few months for the Fable and Mythos line. Fable 5 and Mythos 5 launched June 9, 2026, but access was suspended just three days later to comply with U.S. Department of Commerce export controls, before being restored on July 1 once those controls were lifted. Fable 5.1 arrives on the back of that resolved episode, now positioned as the stable, cost-optimized successor.
For readers tracking the broader AI infrastructure race, TechnoSports’ recent coverage of Apple’s enterprise AI hardware surge is a useful companion read on how demand for both software and hardware AI capability is accelerating in parallel this year.
Bottom Line
Fable 5.1 pushes coding and research benchmarks meaningfully higher while cutting cache costs — a rare combination of “better and cheaper” in the current AI model cycle. Mythos 5.1 remains gated to specialists, keeping Anthropic’s most unrestricted capabilities behind vetted access even as the broader Fable line gets easier and less expensive to use.
