Claude Fable 5 Review: Benchmarks, Pricing & What's New (2026)
On June 9, 2026, Anthropic released Claude Fable 5 — their most capable model available to the public, alongside Claude Mythos 5 for a limited group of cyberdefenders. Here is everything that matters: what it can do, how it benchmarks, who it is for, and the current access situation.
What is Claude Fable 5?
Claude Fable 5 is a Mythos-class model — Anthropic's highest tier — that has been tuned with safeguards to make it safe for general release. Without those safeguards, Fable 5's capabilities in areas like cybersecurity could be misused. Anthropic chose to release it conservatively: a small percentage of sessions (under 5%) are silently routed to Claude Opus 4.8 when the request touches sensitive topics.
Claude Mythos 5 is the same underlying model with those safeguards partially lifted. It is only available to a small group of cyberdefenders and infrastructure providers through Project Glasswing, a US-government collaboration. Anthropic plans to expand the Mythos 5 trusted-access program over time.
Pricing is $10 per million input tokens and $50 per million output tokens — Anthropic notes this is less than half the price of the previous Mythos Preview model.
Key Capabilities
Software Engineering
Stripe reported that Fable 5 compressed months of engineering into days. In a 50-million-line Ruby codebase, the model performed a full migration in a single day that would otherwise have taken a whole engineering team over two months. On Cognition's FrontierCode benchmark — which tests whether models can pass difficult coding tasks while meeting production-code quality standards — Fable 5 scores highest among all frontier models, even at medium effort.
Knowledge Work
On Hebbia's Finance Benchmark for senior-level reasoning, Fable 5 earns the highest score of any model, with substantial gains in document-based reasoning, chart and table interpretation, and problem solving. IMC noted that Fable 5 aced their trading-analysis evaluations nearly across the board, including factual lookup, conceptual reasoning, root-cause analysis, and expected-value analysis.
Vision
Fable 5 is the new state-of-the-art model for vision tasks. It can extract precise numbers from detailed scientific figures and rebuild a web app's source code from screenshots alone — with no additional tooling. Earlier Claude models needed complex helper harnesses to play Pokémon FireRed; Fable 5 completed the game using only raw game screenshots.
Memory and Long-Context
Fable 5 stays focused across millions of tokens in long-running tasks and improves its outputs using its own file-based notes. When tested on the deck-building game Slay the Spire, giving Fable 5 access to persistent memory improved its performance three times more than the same setup did for Opus 4.8. Fable 5 also reached the game's final act three times more often.
Drug Design
Using the Mythos 5 variant, Anthropic's internal protein design experts accelerated aspects of the drug design process by around 10×. In one study, Mythos 5 — equipped with protein design and bioinformatics tools but no human assistance — matched or beat skilled human operators on all tasks: choosing binding sites, selecting and running design tools, and recovering from failures. Nine of 14 protein targets yielded strong drug design candidates currently under investigation.
Fable 5 vs Claude Opus 4.8 — Key Benchmarks
Anthropic published a full benchmark table in the launch announcement. Here are the highlights most relevant for developers and knowledge workers.
| Benchmark | Fable 5 | Opus 4.8 | Note |
|---|---|---|---|
| Cognition FrontierCode | Highest among frontier models | Lower | Passes difficult coding tasks at production-code quality standards, even at medium effort |
| Hebbia Finance Benchmark (senior-level) | Highest of any model | Lower | Document-based reasoning, chart/table interpretation, problem solving |
| Vision — rebuilding apps from screenshots | Yes, no extra tooling | Partial | Fable 5 reconstructs web apps from raw screenshots alone |
| Pokémon FireRed (vision-only harness) | Completed the game | Needed complex helper harness | Fable 5 beat FireRed with a minimal, vision-only harness |
Pricing
Compare to Claude Opus 4.8: $5/M input, $25/M output. Fable 5 costs 2× more per token but delivers meaningfully stronger performance on long-horizon and complex tasks.
Who Is Fable 5 For?
- ✓Large-scale software engineering — Stripe's experience shows it can replace weeks of team effort for codebase migrations
- ✓Finance, legal, and research analysts — top-of-class on document-based reasoning and table/chart interpretation
- ✓Vision-heavy workflows — UI analysis, scientific figure extraction, app reconstruction without extra tooling
- ✓Agentic systems requiring sustained focus — outperforms prior models significantly on tasks spanning millions of tokens
- ✗Cost-sensitive or high-volume API users — Claude Opus 4.8 at half the price is likely the better choice for most production workloads
Frequently Asked Questions
Why is Claude Fable 5 access currently suspended?
On June 12, 2026 — three days after launch — the US government issued an export control directive requiring Anthropic to suspend all access to both Fable 5 and Mythos 5. Anthropic has not specified the exact scope of the restriction but has stated it is working to restore access as quickly as possible. Until access is restored, Claude Opus 4.8 remains the most capable generally available Claude model.
What are the safeguards in Fable 5?
Fable 5 includes conservative content safeguards tuned specifically around cybersecurity misuse risks. On average, fewer than 5% of sessions trigger them — those sessions receive a response from Claude Opus 4.8 instead. Anthropic has said the safeguards were intentionally conservative at launch and will be refined over time to reduce false positives.
What is the difference between Fable 5 and Mythos 5?
Both share the same underlying model weights. Fable 5 includes the full set of safeguards for general use. Mythos 5 has those safeguards lifted in specific areas and is only available to vetted cyberdefenders and infrastructure providers through Anthropic's Project Glasswing program, operated in collaboration with the US government. Mythos 5 has the strongest cybersecurity capabilities of any model commercially available.
Should I use Fable 5 or Claude Opus 4.8?
Access to Fable 5 is currently suspended, so Opus 4.8 is the de-facto best available Claude model right now. When Fable 5 access is restored: for complex, long-horizon tasks (large codebase migrations, multi-document research, vision-heavy work), Fable 5 is clearly the stronger choice. For cost-sensitive production workloads, Opus 4.8 at $5/$25 per million tokens will likely deliver better economics.
Interested in frontier AI models? See the Claude Fable 5 listing or compare it to Claude Opus 4.8.
← Back to all postsAffiliate disclosure: Some links on this page are affiliate links. If you sign up through them, AISO Tools may earn a commission at no extra cost to you. This never affects our rankings or reviews.
📬 Get the best new AI tools delivered weekly
One concise email with fresh launches, trending picks, and featured standouts.
Join thousands of professionals who discover the best AI tools every week. No spam — unsubscribe anytime.