Claude Fable 5

An independent benchmark and specifications recap confirms Claude Fable 5 was released on June 9, 2026, as the first Mythos-class model Anthropic has made generally available, posting state-of-the-art results across software engineering, knowledge work, scientific research, and vision benchmarks. The API model ID is cl Fable 5 is described as "Mythos made safe," with a separate classifier system monitoring every session for high-risk cybersecurity, biology, chemistry, and distillation requests and rerouting flagged requests to a category-appropriate fallback model such as Opus 4.8 for cybersecurity or Opus 5 for biology, with the use

Explainx.ai's September 2, 2026 breakdown of the September 1-2 joint release covers Claude Fable 5.1 as the generally available model and Claude Mythos 5.1 as the same weights with select safeguards lifted for vetted cybersecurity and life-sciences users via Project Glasswing. It reports headline benchmarks including T On pricing, input/output remain $10/$50 per million tokens while cache reads drop 75% from $1.00 to $0.25, yielding an estimated 25% cost reduction for typical workloads and up to 45% for highly agentic work. A new Enterprise Frontier Safeguards (EFS) feature with customer-cloud data retention is rolling out in fall 20

Anthropic's official Claude Fable product page documents the model's evolution across 2026: the June 9 launch of Claude Fable 5 (5th generation for the most ambitious work, handling days-long complex tasks), a June 12 access disruption, a July 1 access restoration, an August 6 biology-safeguards update reducing false p The page specifies Fable 5.1 availability on Pro, Max, Team, and Enterprise plans, plus the Claude Platform, AWS, Google Cloud, and Microsoft Foundry, with US-only inference available at 1.1x pricing. API pricing is $10 per million input tokens and $50 per million output tokens, with cache reads cut 75% to $0.25 per mi

This kie.ai analysis covers Claude Fable 5.1, which Anthropic officially launched on September 1, 2026 alongside Claude Mythos 5.1, positioned for coding and knowledge work. Confirmed headline pricing is reported as $10 per million input tokens and $50 per million output tokens, with cache reads falling 75% from Fable The piece functions as a retrospective on pre-launch speculation versus what actually shipped, noting that community discussion, account testing reports, and support-document sightings correctly pointed toward unchanged headline pricing and a focus on long-horizon reasoning and agent work, but were not the product docu

On June 9, 2026, Anthropic publicly released Claude Fable 5 as its first generally available model from a new Mythos-class tier that sits above Opus, while keeping Claude Mythos 5 restricted to approved cyberdefense and infrastructure partners via Project Glasswing. According to the Coursiv aggregator citing Anthropic Fable 5 and Mythos 5 share the same underlying model, with Fable 5 adding stricter safety classifiers in cybersecurity, biology, chemistry, and model distillation to make it safe for broad release. Adaptive thinking is always on with summarized rather than raw chain-of-thought output, and best-use framing targets auton

This third-party explainer covers Anthropic's June 9, 2026 launch of Claude Fable 5 alongside Claude Mythos 5, both positioned in a newly introduced Mythos-class tier that sits above Opus in capability. The article describes Fable 5 and Mythos 5 as Anthropic's most capable models to date, built for long-horizon agentic The article situates Claude Fable 5 within Anthropic's naming history, noting the Opus/Sonnet/Haiku structure dates to the Claude 3 generation in March 2024 and that Mythos-class is the first new capability tier added above Opus since then. It emphasizes that Fable 5 and Mythos 5 represent a qualitative rather than inc

On June 9, 2026, Anthropic made Claude Fable 5 generally available, marking the first public release from its new Mythos capability tier, according to a June 12, 2026 GMI Cloud blog post. The piece situates Fable 5 as the outcome of a deliberate post-Opus evolution rather than an incremental upgrade, describing an arc The same post frames Fable 5 against the backdrop of Anthropic's Project Glasswing safety program, noting that a Mythos Preview was limited in April 2026 to cyber defenders and critical infrastructure providers before access expanded to hundreds of organizations across 15 countries and finally reached general availabil

Towards AI's coverage describes Claude Fable 5 as Anthropic's democratized Mythos-class intelligence engine, released on June 9, 2026 at roughly 2x the price of current Opus models and less than half the price of the original Mythos Preview. It emphasizes that Fable 5 ships with a responsible safety layer on top of cap On SWE-Bench Pro, Fable 5 posts 80.3%, an 11-point lead over Anthropic's Opus 4.8 at 69.2%, with GPT-5.5 at 58.6% and Gemini 3.1 Pro at 54.2%. Towards AI positions Fable 5 as genuinely redefining what general-purpose AI can accomplish at this capability tier, though the piece is third-party commentary restating Anthrop

Labellerr's June 10, 2026 review confirms that Claude Fable 5 and Claude Mythos 5 were both released on June 9, 2026, sharing the same underlying architecture and sitting above Opus in a new "Mythos-class" tier. Fable 5 is the general-availability variant, while Mythos 5 is restricted to a small set of approved Project The article details Fable 5's three layers of safety classifiers—cybersecurity, biology and chemistry, and others—running on top of the shared base model, with internal evaluations and a 1,000+ hour external bug-bounty program reporting no universal jailbreaks. When classifiers trigger on Fable 5, responses are automat

StartupFeed's coverage reports that Anthropic launched Claude Fable 5 on June 9, 2026 as its most capable public AI model and the first public release of the Mythos-class tier, which sits above Claude Opus. The article documents API pricing of $10 per million input tokens (roughly Rs 840) and $50 per million output tok It also explains a notable behavioral detail: on certain high-risk topics, Claude Fable 5 queries are silently routed to a weaker model (Opus 4.8) via new safety classifiers, with Claude Mythos 5 shipping to vetted partners on the same day. The piece includes Indian SaaS and developer-tool framing with an editorial rea

Coursiv's guide documents concrete Claude Fable 5 API specifics: model ID `claude-fable-5`, a 1M-token context window, up to 128k output tokens, and pricing of $10 per million input tokens and $50 per million output tokens, with an adaptive "thinking" mode always on. The model is positioned as Anthropic's first general Coursiv reiterates that Fable 5 shares its underlying weights with Claude Mythos 5 but adds stricter safety classifiers and routing/fallback behavior for high-risk areas like cybersecurity, biology, chemistry, and model distillation. It also confirms the June 9, 2026 launch, the June 12 suspension, the July 1, 2026 glo

Kingy AI's analysis confirms Claude Fable 5 launched on June 9, 2026 as Anthropic's first generally available Mythos-class model, a tier above Opus, with Mythos 5 reserved for Project Glasswing and approved cybersecurity partners. The two share the same underlying model, but Fable 5 ships with additional safeguards tha The page highlights headline coding and agentic benchmarks: 80.3% on SWE-bench Pro, 95.0% on SWE-bench Verified (per Vals), 29.3% on FrontierCode Diamond, 72.9% on CursorBench 3.1 at max effort, and 88.0% on Terminal-Bench 2.1, noting that some starred results reflect Mythos-level capability more than the fully safegua