--- title: "Anthropic's Claude Fable 5 Beats Every Frontier Model on Coding Benchmarks" subtitle: "Anthropic's Mythos-class Claude Fable 5 is now publicly available in Zed, bringing frontier-grade agentic coding to your editor for the first time." url: "https://alphasignal.ai/news/anthropic-s-claude-fable-5-beats-every-frontier-model-on-coding-benchmarks" author: "AlphaSignal Newsroom" author_url: "https://alphasignal.ai/editorial-team" company: "Zed" company_url: "https://alphasignal.ai/company/zeddotdev" published: "2026-06-09T21:26:56.000Z" updated: "2026-06-10T00:06:36.594Z" type: "News" topics: - "Development" - "Api" subtopics: - "Code Generation" upvotes: 356 read_time: "6 min" source_url: "https://x.com/zeddotdev/status/2064459254740705423" comments: 0 --- # Anthropic's Claude Fable 5 Beats Every Frontier Model on Coding Benchmarks > Anthropic's Mythos-class Claude Fable 5 is now publicly available in Zed, bringing frontier-grade agentic coding to your editor for the first time. Anthropic just crossed a threshold it had been carefully guarding since April. [Claude Fable 5](https://www.anthropic.com/news/claude-fable-5-mythos-5) -- the first Mythos-class model available to the general public -- is now live, and the [Zed editor](https://zed.dev) is one of the first places you can reach for it, available in both Zed v1.5.5 (stable) and v1.6.2 (preview). This is not a routine model refresh. Mythos was initially limited to a handful of partners due to cybersecurity concerns when it launched in April. What changed is that Anthropic spent the intervening months building classifiers strong enough to make a public release defensible. The result is Fable 5: same underlying model, with hard guardrails on the most dangerous capabilities. ## A tier above Opus Fable 5 is a Mythos-class model made safe for general use, and its capabilities exceed those of any model Anthropic has ever made generally available. It is state-of-the-art on nearly all tested benchmarks, showing exceptional performance in software engineering, knowledge work, vision, and scientific research. The longer and more complex the task, the larger Fable 5's lead over other models. That framing is backed by numbers: - Fable 5 reaches 95.0% on SWE-bench Verified and 80.0% on SWE-bench Pro, compared to Opus 4.8 at 88.6% and 69.2% respectively. - On SWE-Bench Pro, Fable 5 scored 80.3%, ahead of GPT-5.5 at 58.6% and Gemini 3.1 Pro at 54.2%. On FrontierCode Diamond, it scored 29.3% compared to 13.4% for Opus 4.8 and just 5.7% for GPT-5.5. - During early testing, Stripe reported that Fable 5 compressed months of engineering into days -- performing a codebase-wide migration in a 50-million-line Ruby codebase in a single day, a task that would otherwise have taken a whole team over two months. - On Cognition's FrontierCode evaluation, which tests whether models can pass difficult coding tasks while meeting the standards of high-quality production codebases, Fable 5 scores highest among frontier models, even at medium effort. ## What it can actually do Run Fable 5 in an agent harness like Claude Code or Claude Managed Agents, and it can work for days at a time: planning across stages, delegating to sub-agents, and checking its own work. This is the key shift. Previous models could handle a task. Fable 5 is designed to handle a project. Beyond code, the model brings meaningful upgrades across several domains: - **Vision:** It can write its own tests to check its work, implement designs with high fidelity, and use vision to check outputs against goals. - **Knowledge work:** Fable 5 handles complex, multi-stage knowledge work with minimal oversight -- from deep research and analysis to deliverables ready for review. Teams can hand off large projects and review completed work rather than supervising every step. - **Memory:** The model stays focused across millions of tokens in long-running tasks and improves its outputs using its own notes -- a capability that benefits significantly from persistent file-based memory in agentic setups. - **Finance and analytics:** On Hebbia's Finance Benchmark for senior-level reasoning, Fable 5 has the highest score of any model, with substantial gains in document-based reasoning, chart and table interpretation, and problem solving. ## The safety architecture that made this possible The reason Mythos stayed locked up for two months is that without safeguards, Fable 5's capabilities in areas like cybersecurity could be misused to cause serious damage. Anthropic's solution is a layered classifier system -- separate AI models that intercept requests before Fable 5 ever sees them. Queries on flagged topics are routed to Claude Opus 4.8 instead. The safeguards are tuned conservatively and will sometimes catch harmless requests, though they trigger, on average, in less than 5% of sessions. Covered domains include cybersecurity, biology, chemistry, and attempts to distill the model's capabilities to train competing systems. There is one notable tradeoff that enterprise users need to know about. Using Fable requires 30-day data retention for safety monitoring. Retained data is not used to train Anthropic's models , but it does mean that previous zero-data-retention agreements no longer apply to Fable 5 traffic. You won't be charged Fable prices for rerouted requests. The red-teaming story is also worth noting. Anthropic stress-tested its classifiers with jailbreak attempts before release, running an external bug bounty that produced no universal jailbreaks in over 1,000 hours of testing. External red-teaming organizations also failed to find universal jailbreaks. ## Fable vs. Mythos: same model, different keys For a small group of cyberdefenders and infrastructure providers, Anthropic is also launching Claude Mythos 5. It's the same underlying model as Fable 5, but with the safeguards lifted in some areas. Mythos 5 will initially be deployed through Project Glasswing, in collaboration with the US government. The naming is deliberate. Fable comes from the Latin *fabula* -- "that which is told" -- akin to the Greek *mythos*. The safeguards are what distinguish the two products, not the weights. ## Pricing and access Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens -- less than half the price of Claude Mythos Preview. The rollout for subscription plans is staged: - 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, Fable 5 will be removed from those plans. Using it after that will require usage credits. - After this point, when sufficient capacity allows, Anthropic aims to restore Fable 5 as a standard part of subscription plans. For developers building AI solutions, Claude Fable 5 is available on the Claude Platform natively, through available marketplaces, and in Amazon Web Services, Google Cloud, and Microsoft Foundry. The API model ID is `claude-fable-5`. ## What this means for the field The Fable 5 launch reframes what "publicly available" means for frontier AI. With the launch of Claude Fable 5, Anthropic is honoring its stated goal to deploy Mythos-class models at scale. The mechanism it used to get there -- a classifier layer that degrades gracefully to a strong fallback model rather than refusing outright -- is likely to become the template for how labs release increasingly capable systems going forward. In a telling convergence, both Anthropic and OpenAI now treat cybersecurity and biology as high-risk and gate those capabilities behind vetted-access programs. The question of how much capability to expose publicly, and under what conditions, is no longer a theoretical one. Fable 5 is the first concrete answer at this capability tier -- and it landed in your code editor today. ## Sources - [X post by @zeddotdev](https://x.com/zeddotdev/status/2064459254740705423) - [Claude Fable 5 and Claude Mythos 5 — Anthropic](https://www.anthropic.com/news/claude-fable-5-mythos-5) - [Anthropic releases Claude Fable, a version of Mythos — TechCrunch](https://techcrunch.com/2026/06/09/anthropic-released-claude-fable-5-its-most-powerful-model-publicly-days-after-warning-ai-is-getting-too-dangerous/) - [Claude Fable 5 product page — Anthropic](https://www.anthropic.com/claude/fable) - [Claude Fable 5 is generally available for GitHub Copilot — GitHub Changelog](https://github.blog/changelog/2026-06-09-claude-fable-5-is-generally-available-for-github-copilot/) - [Claude Fable 5 Benchmark Scores — Weights & Biases](https://wandb.ai/byyoung3/ml-news/reports/Claude-Fable-5-Benchmark-Scores--VmlldzoxNzE3NTE3MQ)