Claude Fable 5 Is Out: Anthropic's First Public Mythos Model
June 9, 2026
Anthropic just did something it hasn't done before. It shipped a single frontier model as two distinct products — one for everyone, one for vetted partners only.
Claude Fable 5 (claude-fable-5) is the new generally available flagship, wrapped in safety classifiers that hand certain queries off to Claude Opus 4.8. Claude Mythos 5 is the same underlying model with those safeguards lifted in some areas, restricted to vetted partners working on cyberdefense and infrastructure.
Both launched today, June 9, 2026. If you've been watching the Project Glasswing rollout since April, this is the public chapter.
What Fable 5 Actually Is
Anthropic describes Fable 5 as a Mythos-class model made safe for general use — its highest-capability model released publicly to date. It performs at state-of-the-art levels on nearly all tested benchmarks across software engineering, knowledge work, vision, and scientific research. Its advantage over previous models increases with task length and complexity.
The benchmark numbers back that up. Fable 5 hits 80.3% on SWE-Bench Pro — Opus 4.8 scores 69.2%, GPT-5.5 scores 58.6%. It's also the first model to exceed 90% on Hex's analytical benchmark.
On the harder FrontierCode Diamond set, Fable 5 scores 29.3% against 13.4% for Opus 4.8 — more than double in relative terms. Knowledge work, spatial reasoning, tool use, legal, and health benchmarks show the same shape: a clear lead over Opus 4.8 and everything else currently available.
One caveat: starred benchmarks show a larger gap between Fable 5 and Mythos 5 due to blocking safeguards for cybersecurity and biology-related queries. On those, Fable 5 performs closer to Opus 4.8 because of fallbacks. More on that below.
The Safeguard Split: What Gets Routed to Opus 4.8
This is the most important thing to understand about Fable 5 before you build on it.
The model includes new safeguards that redirect certain queries to Claude Opus 4.8, the next-most-capable model. Specifically: cybersecurity, biology, and model distillation queries trigger the fallback.
Both Anthropic and OpenAI have converged on gating cyber and biology. If your work lives near those domains, expect more refusals or fallbacks from both — and a vetted-access path from both.
For most developers building on coding, reasoning, knowledge work, or general-purpose agents — you won't hit these limits. If you're working on security tooling, expect the same friction you'd have seen in Mythos Preview, and look at the Project Glasswing access path instead.
Pricing
Fable 5 is priced at $10 per million input tokens and $50 per million output tokens, with the existing 90% input token discount for prompt caching.
That's a significant reduction from the $25 and $125 per million token pricing that applied to the restricted Mythos Preview since April. It's double Opus 4.8 on absolute terms, but given the benchmark gap — especially on long, complex tasks — the cost-per-unit-of-work math is worth running for your use case before assuming it's too expensive.
For comparison: GPT-5.5 sits at $5/$25 per million tokens. Fable 5 wins the benchmarks. GPT-5.5 wins the price. Which one fits depends on the work.
How to Access It
Fable 5 is available immediately on the Claude API (model ID claude-fable-5), Claude Platform, Claude Code, consumption-based Enterprise plans, and on AWS, Google Cloud, and Microsoft Foundry.
Subscription plan users: there's a window to be aware of. From today through June 22, Fable 5 is included on Pro, Max, Team, and seat-based Enterprise plans at no extra cost. On June 23, Anthropic will remove it from those plans — using it after that will require usage credits. Anthropic says it aims to restore Fable 5 as a standard part of subscription plans when sufficient capacity allows.
If you want to try it on a subscription plan, do it this week.
What About Mythos 5?
Alongside Fable 5, Anthropic revealed Claude Mythos 5 — the same underlying model with fewer safeguards. Beginning today, all users who currently have access to Claude Mythos Preview through Project Glasswing are being upgraded to Mythos 5.
Anthropic plans to expand access to Mythos 5 through a broader trusted access program in the coming weeks. If you're doing defensive security work and haven't applied to Glasswing yet, now is the time.
What This Means for Your Stack
If you're on Opus 4.8 today, Fable 5 is a meaningful upgrade for any workload involving long, complex, multi-step tasks — especially agentic coding. The SWE-Bench gap is significant and it widens further on harder problems.
For teams on tighter budgets running high-volume, shorter tasks: Opus 4.8 still makes sense. The price difference is real and Opus 4.8 is a strong model.
The June 15 deprecation of claude-sonnet-4-0 and claude-opus-4-0 is still live — if you haven't migrated those yet, now's the time to skip to Fable 5 directly rather than landing on an intermediate model.
Sources: Anthropic, Digital Applied, Investing.com