Claude Fable 5: Anthropic’s Mythos-Class Model Is Now Open to Everyone
Anthropic launched claude fable 5 — a Mythos-class model made safe for general use, with capabilities that exceed those of any model the company has ever made publicly available. That is not a small claim. On SWE-Bench Pro, Fable 5 scored 80.3%, roughly 11 points ahead of the next-best frontier model — and the gap between Fable 5 and Anthropic’s own Opus 4.8 is larger than the gap between Opus 4.8 and Google’s Gemini 3.1 Pro. Something genuinely new just arrived at the frontier.
This article breaks down everything: what claude fable 5 actually is, how it performs on benchmarks, what it costs, why the safeguards matter, and how it fits into the broader arc of autonomous AI coding agents reshaping enterprise software.
What Is Claude Fable 5? Understanding the Mythos Class
The Claude family has covered three named size tiers — Opus, Sonnet, and Haiku — since March 2024. In June 2026, a Mythos-class tier that sits above Opus was introduced, with the current frontier being Claude Fable 5, announced on June 9, 2026.
Claude Fable 5 is part of a model family that now spans four classes: Haiku, Sonnet, Opus, and Mythos. So when Anthropic says this is a mythos class ai model, they mean it sits at the very top of their capability hierarchy — not just an incremental bump.
Anthropic released two new fifth-generation AI models: Claude Fable 5 for general use, and Claude Mythos 5, which is initially available only to selected partners for specialized areas such as cybersecurity. The key distinction is access. Same core engine. Different safeguard layers. Claude Fable 5 and Claude Mythos 5 use the same underlying model. Fable 5 has safeguards for general use. Mythos 5 has some safeguards lifted and remains limited to approved trusted-access programs.
The frontier crossed a notable threshold in April 2026 with Mythos Preview, when Anthropic’s most capable model found thousands of zero-day vulnerabilities autonomously, including 27-year-old OpenBSD bugs — deemed too powerful for general release, shipping only through Project Glasswing to critical-infrastructure partners at $25/$125 per million tokens. Now, months later, a public version is here. When Anthropic first released Mythos Preview, the company stated it hoped to eventually release Mythos-level capabilities to all users, so long as new safeguards were strong enough to reliably prevent misuse. Over the past few months, those safeguards have been improved to be robust enough for a general release.
Claude Fable 5 Benchmarks: A Performance Leap That’s Hard to Ignore
The claude fable 5 benchmarks tell a clear story: this is the biggest coding performance jump at the frontier in recent memory.
Claude Fable 5 leads with a performance of 95.00% on SWE-bench Verified, achieving the best accuracy on that benchmark. Claude Opus 4.8 follows at 88.60%. GPT-5.5 comes in third with 82.60%, followed by Claude Opus 4.7 at 82.00%.
On harder, less saturated benchmarks, the story gets even more interesting:
On SWE-Bench Pro, Claude Fable 5 posts 80.3% versus 69.2% for Opus 4.8, 58.6% for GPT-5.5, and 54.2% for Gemini 3.1 Pro.
Anthropic’s launch chart reports 29.3% for Claude Fable 5 on FrontierCode Diamond, compared with 13.4% for Claude Opus 4.8 and 5.7% for GPT-5.5.
Claude Fable 5 also scored 88.0% on Terminal-Bench 2.1 and 59.0% on Humanity’s Last Exam without tools.
On HealthBench Professional, Claude Fable 5 scored 66.0%, ahead of Claude Opus 4.8 at 56.9% and GPT-5.5 at 51.8%.
The biggest jump shows up on long-horizon agentic coding benchmarks where a model must investigate, patch, test, and recover over many steps. Teams seeing the best outcomes are giving Fable 5 harder, previously unsolved problems — the benchmark pattern confirms that Fable separates most clearly when the work requires persistence.
On ViBench, an end-to-end vibe-coding benchmark, Claude Fable 5 is the highest-performing model tested — nearly saturating base use cases and building apps in less time with fewer tokens. It also beats Opus 4.8 on everyday spreadsheet tasks at every effort level, finishing runs 25–30% faster.
Real-world enterprise results reinforce the benchmark data. Stripe reported that Fable 5 completed a codebase-wide migration across a 50-million-line Ruby codebase in a day — a task Anthropic says would otherwise have taken a full team more than two months. That is not a benchmark. That is production output, compressed by orders of magnitude.
Analytics company Hex said Fable was the first to get a 90% on its core analytics benchmark of complex, long-running analytical tasks, noting “on the hardest questions, it shows strong judgement and attention to nuance.” AI-powered workspace platform Genspark said Fable beat every other model in its evaluations and performed significantly better on tasks like UI design and game coding.
Claude Fable 5 Pricing: What It Costs and Who It’s For
Let’s talk numbers, because they matter — especially for teams evaluating whether to upgrade.
Pricing for both Fable 5 and Mythos 5 is $10 per million input tokens and $50 per million output tokens, double the price of Opus 4.8. This cuts Mythos Preview pricing by more than 50%, which is a meaningful shift for the Mythos tier. But it is still steep compared to earlier Claude models.
Official claude fable 5 pricing is $10 per million input tokens and $50 per million output tokens. Batch pricing is $5 per million input tokens and $25 per million output tokens. The batch option could make high-volume workflows significantly more economical for engineering teams running overnight jobs.
Fable 5 is available now through the Claude API and usage-based Enterprise plans. Subscription plans — Pro, Max, Team, and seat-based Enterprise — follow a staggered rollout. Until June 22, Fable 5 is included at no extra cost. Starting June 23, access requires usage credits.
Many enterprises are growing critical of AI costs after seeing bills come in or blowing through yearly AI budgets early. Advanced models like Opus 4.8 can exacerbate those issues, with advanced reasoning skills that can split a single request into multiple tasks. Fable 5 is twice as expensive again. That makes ROI framing essential.
Anthropic says Fable 5 can work unattended for longer and with more independence than previous Claude models, which is exactly the capability enterprises need if they want AI agents to do more than autocomplete code or answer developer questions. The calculation, then, is not cost-per-token but value-per-project.
Claude Fable 5 Safeguards: How Anthropic Made Mythos Safe for Public Use
This is where things get nuanced. The claude fable 5 safeguards are not optional — they are structural.
Anthropic says Fable 5 uses Mythos-level technology but is safe for general use because it has added limits and safeguards. Sensitive prompts tied to cybersecurity, biology, and chemistry will be routed to Opus 4.8, its previous top public model.
To release the model both safely and quickly, Anthropic tuned these safeguards conservatively — they will sometimes catch harmless requests, though they trigger, on average, in less than 5% of sessions. That 5% figure is low enough not to disrupt most workflows but high enough to feel occasionally restrictive if your work touches regulated domains.
Internal and external security researchers spent more than 1,000 hours attempting to bypass the model’s protections. Testers failed to discover a universal method that could consistently defeat the safety systems.
With the launch of Fable 5 and Mythos 5, Anthropic requires a 30-day retention on all traffic, even if enterprises previously had zero-retention agreements. The company says it won’t use the data for training and will use it only to defend against complex and novel attacks, including new jailbreaks, and to identify and reduce false positives.
That policy may become one of the most important enterprise buying questions around Fable 5. Many businesses want frontier AI capability but also want strict control over data retention, especially in regulated sectors. Anthropic’s position is that stronger monitoring is necessary for models with this level of capability. Enterprise customers will have to decide whether the capability gain justifies the retention requirement.
Anthropic says Mythos-class models are powerful enough to pose real risks, in cyberattacks or bioweapons research, for example. The stakes are unusually high. That context makes the conservative safeguard design feel deliberate, not overcautious.
Claude Mythos 5 Release: The Restricted Twin
The claude mythos 5 release happened simultaneously — and it’s a different story entirely.
In April 2026, Anthropic began Project Glasswing, releasing the first Mythos-class model — Claude Mythos Preview — to only a limited group of cyber defenders and critical software infrastructure providers. Last week before the Fable 5 launch, Anthropic expanded access to hundreds of organizations across 15 countries, again focusing on organizations that manage critical infrastructure.
Claude Mythos 5, with reduced safety restrictions and shared architecture, is available via “Project Glasswing” to approximately 200 qualified institutions for cybersecurity applications.
In drug design using Mythos 5, internal protein design experts accelerated aspects of the drug design process by around ten times. Mythos 5, with protein design and bioinformatics tools but no human assistance, matches or beats skilled human operators — executing all of the tasks that are normally completed by a scientist: choosing binding sites, selecting and running protein design tools, and recovering from failures along the way.
The distinction is clean: Fable 5 for the world, Mythos 5 for vetted specialists. Same underlying model, radically different access conditions.
Autonomous AI Coding Agents: Why Fable 5 Changes the Game
The rise of autonomous ai coding agents is the real business story underneath these benchmarks. Claude Fable 5 is the first model in Anthropic’s Mythos class, designed for long-horizon, autonomous coding and knowledge-work tasks.
Fable is for the most ambitious, asynchronous tasks — big projects you want the model to break down, research, create, and verify on its own over extended periods. That repositions autonomous ai coding agents from a novelty to a genuine engineering resource.
Zapier said Fable 5 is the new leader on AutomationBench and is more autonomous than Opus 4.8: “Where Opus stops to ask, Fable 5 keeps looking.” For enterprise software vendors, this points toward more capable embedded agents in workflow products — ones that can chain together multiple complex actions without intervention.
In Anthropic’s automated alignment assessment, Mythos 5’s level of misaligned behavior — including deception and cooperation with misuse — was low, and similar to that of Opus 4.8. Given they are the same underlying model, Fable 5’s alignment will be similar. For teams deploying AI agents in production, that alignment parity with the well-tested Opus 4.8 is genuinely reassuring.
Claude Fable 5 is now available on the Databricks Lakehouse Platform. This rollout brings the model’s capabilities for autonomous knowledge work directly to enterprise data. Anthropic’s 5th-generation AI model makes Mythos-level capabilities available to all customers and shows exceptional performance in coding, knowledge work, and vision — now available on Amazon Bedrock and Claude Platform on AWS.
Where and How to Access Claude Fable 5
Claude Fable 5 is now available to users through Claude products and the Claude API. Developers can access the model through the claude-fable-5 API identifier.
Claude Fable 5 is now available in GitHub Copilot — the first model in Anthropic’s Mythos class designed for long-horizon, autonomous coding and knowledge-work tasks. Unlike other Claude models in GitHub Copilot, Claude Fable 5 requires data retention to operate Anthropic’s safety classifiers.
Through June 22, Fable 5 is included in Pro, Max, Team, and seat-based Enterprise plans at no extra cost. Starting June 23, access on subscriptions requires usage credits. Down the line, Anthropic plans to fold the model back into regular subscription plans once it has enough capacity.
For API builders: Fable 5’s safeguards route flagged cybersecurity, biology, chemistry, and distillation queries to Opus 4.8 — under 5% of sessions — and users are not charged Fable prices for those routed responses. That pricing detail is easy to miss and worth knowing before scaling a deployment.
The Bigger Picture: What Fable 5 Means for the AI Landscape
Fable’s launch comes as Anthropic prepares to enter the public markets. It also follows the AI firm’s plea urging major global AI labs to establish a coordinated brake pedal on frontier AI development. Anthropic warned that systems are advancing so rapidly that they may soon achieve recursive self-improvement — autonomously improving themselves without human intervention.
That context makes the claude mythos 5 release and Fable 5’s safeguard architecture feel less like product decisions and more like a policy position. Anthropic is making a public argument: that frontier capability and responsible deployment can coexist — if you build the infrastructure for it.
Anthropic’s revenue trajectory alone tells a striking story: from $1 billion ARR in December 2024 to $9 billion by end of 2025, $14 billion in February 2026, $30 billion by April, and $47 billion by late May 2026. The company is not navigating this question from a position of weakness. It is setting norms from the top of the market.
Whether you are an engineer evaluating claude fable 5 benchmarks for your next sprint, a decision-maker weighing claude fable 5 pricing against your AI budget, or a researcher tracking the frontier, the message from this launch is the same: the Mythos era has begun, and it is now public.
Frequently Asked Questions
What is Claude Fable 5?
Claude Fable 5 is Anthropic’s first Mythos-class model made safe for general use. Its capabilities exceed those of any model Anthropic has ever made generally available, achieving state-of-the-art results on nearly all tested benchmarks of AI capability, including exceptional performance in software engineering, knowledge work, vision, and scientific research.
When was Claude Fable 5 released?
Claude Fable 5 was released on June 9, 2026, making it the newest generally available Claude model. Claude Mythos 5 launched on the same day with restricted access.
What are the Claude Fable 5 benchmarks?
Fable 5 scored 95.0% on SWE-bench Verified, 80.0% on SWE-bench Pro, and 72.9% on CursorBench at max effort, and led FrontierCode in both Diamond and Main subsets. On HealthBench Professional, it scored 66.0%, ahead of Opus 4.8 at 56.9%.
What is Claude Fable 5 pricing?
Anthropic has priced both Fable 5 and Mythos 5 at $10 per million input tokens and $50 per million output tokens — roughly double the cost of Claude Opus 4.8. Batch pricing is $5 per million input tokens and $25 per million output tokens.
What are the Claude Fable 5 safeguards?
For high-risk queries, Fable 5 features built-in rigorous safety protections that block queries in high-risk areas like cybersecurity and biology, automatically switching to the Opus 4.8 model to generate a safe response when such issues are encountered. To prevent models at the Mythos level from falling into the wrong hands, Anthropic conducted over 1,000 hours of internal and external red-teaming tests.
How is Claude Fable 5 different from Claude Mythos 5?
Fable 5 and Mythos 5 share the same underlying model. Fable 5 is the safe-for-general-use release; Mythos 5 lifts certain safeguards and is restricted to vetted partners. Fable 5 demonstrates over 10% performance improvement in software engineering and knowledge work tasks compared to its predecessor.
Where can I access Claude Fable 5?
Claude Fable 5 is available through Claude products and the Claude API, accessible via the claude-fable-5 API identifier.It is also available on Amazon Bedrock and Claude Platform on AWS, in GitHub Copilot, and on the Databricks Lakehouse Platform.