Anthropic Launches Claude Fable 5 and Claude Mythos 5: A New Frontier in AI Capability

Anthropic has released two landmark AI models - Claude Fable 5 and Claude Mythos 5 - on June 9, 2026, marking the first time a Mythos-class model has been made available for general use. Fable 5 is now accessible via the public API and major cloud platforms, while Mythos 5, the same underlying model with cybersecurity safeguards lifted, remains restricted to vetted partners under Project Glasswing. Both models represent the most capable AI systems Anthropic has ever deployed, with exceptional performance across software engineering, scientific research, vision, and cybersecurity.

Quick Facts

  • Release Date: June 9, 2026
  • Developer: Anthropic
  • Model Names: Claude Fable 5 (public) and Claude Mythos 5 (restricted)
  • Pricing: $10 per million input tokens and $50 per million output tokens - less than half the cost of Claude Mythos Preview
  • Context Window: 1 million input tokens, up to 128,000 output tokens
  • API Model ID: claude-fable-5
  • Availability: Claude API, AWS Bedrock, Vertex AI, and Microsoft Foundry (Fable 5); Project Glasswing partners only (Mythos 5)
  • Subscription Plans: Fable 5 included at no extra cost through June 22, 2026, for Pro, Max, Team, and seat-based Enterprise plans
  • Project Glasswing: Expanded to approximately 150 new organizations across more than 15 countries as of June 2, 2026

What Happened?

On June 9, 2026, Anthropic officially announced and released Claude Fable 5 and Claude Mythos 5. Fable 5 is described by Anthropic as a Mythos-class model made safe for general use, featuring capabilities that exceed any model the company has previously made publicly available. The name "Fable" is derived from the Latin word "fabula" - meaning "that which is told" - itself related to the Greek word "mythos." The two names reflect the distinction in safeguards rather than in underlying capability: Fable 5 and Mythos 5 share the same base model, but differ in how safety classifiers are applied.

Claude Mythos 5 is an upgrade to Claude Mythos Preview, which was first launched in April 2026 under Project Glasswing - an invitation-only program for vetted cybersecurity and infrastructure partners. Prior to the June 9 release, Mythos Preview had been available to approximately 50 founding partner organizations. On June 2, 2026, Anthropic expanded Project Glasswing to roughly 150 additional organizations spanning power, water, healthcare, communications, and hardware sectors. The June 9 launch of Mythos 5 replaces Mythos Preview for all current Glasswing partners.

Key Facts

  • Fable 5 is the first publicly available Mythos-class model and Anthropic's most capable general-release AI system to date.
  • Mythos 5 is the same underlying model as Fable 5, but with cybersecurity and other safety classifiers lifted for vetted partners.
  • Fable 5's safety safeguards route certain sensitive queries to Claude Opus 4.8 as a fallback, triggered on average in fewer than 5% of sessions.
  • The API returns HTTP 200 with a stop_reason of "refusal" when safeguards are triggered, naming the category - "cyber," "bio," or "reasoning_extraction."
  • A beta "fallbacks" parameter allows developers to retry refused requests server-side in a single round trip (available on Claude API and Claude Platform on AWS only).
  • Fable 5 scores state-of-the-art results on nearly all tested AI capability benchmarks, including software engineering, knowledge work, vision, and scientific research.
  • Mythos 5 is described by Anthropic as having the strongest cybersecurity capabilities of any model in the world.
  • In life sciences testing, Mythos 5 accelerated aspects of the drug design process by approximately ten times, with nine out of 14 protein targets yielding strong drug design candidates.
  • Mythos 5 conducted autonomous genomics research, assembling single-cell data for millions of cells across 138 animal species and training a custom machine learning model that outperformed a recent model published in the journal Science, despite being 100 times smaller.
  • Early adopter Stripe reported that Fable 5 performed a codebase-wide migration of a 50-million-line Ruby codebase in a single day - a task that would have taken a full engineering team over two months.
  • Anthropic is committing up to $100 million in usage credits for Project Glasswing partners, per reporting from the Centre for Emerging Technology and Security at the Alan Turing Institute.
  • The existence of the Mythos model was first publicly revealed on March 26, 2026, following a data configuration incident that exposed draft blog posts and internal documents.

Why It Matters

The launch of Claude Fable 5 and Mythos 5 represents a significant milestone in the public deployment of frontier AI. For the first time, Anthropic has made a Mythos-class model - previously restricted entirely to vetted security partners - available to the general public via API and consumer applications. This marks a qualitative shift from the incremental Haiku-Sonnet-Opus release pattern that Anthropic has followed since Claude 3.

The decision to release a model of this capability level also introduces new questions about AI safety at deployment scale. Anthropic has acknowledged that, without safeguards, Fable 5's capabilities in areas such as cybersecurity could be misused to cause serious damage. The company's approach of routing sensitive queries to a less capable fallback model - rather than blocking them outright - is a novel safety design choice with implications for how other AI developers may handle similar situations in the future.

Independent scrutiny of Mythos-class capabilities has also emerged. Researchers have questioned whether some demonstrated cybersecurity achievements reflect genuine novel AI capability or effective orchestration using existing models and known vulnerability data. Niels Provos, a credited contributor to historical BSD security work, reportedly reproduced certain findings using smaller open-weight models, concluding that orchestration infrastructure - not frontier model capability alone - may account for much of the observed performance, according to reporting by flyingpenguin.com.

What It Means for India

No verified information available from the source for this section.

Industry Impact

The release of Fable 5 has immediate implications for the software engineering and enterprise AI sectors. Stripe's reported experience - compressing months of engineering work into a single day - illustrates the potential productivity transformation for large-scale software organizations. Financial analytics firm IMC reported that Fable 5 performed at or near perfect scores across their trading-analysis evaluations, including factual lookup, conceptual reasoning, root-cause analysis, and expected-value analysis, per Anthropic's announcement.

In cybersecurity, Project Glasswing's expansion to critical infrastructure sectors - power, water, healthcare, and communications - signals Anthropic's intent to position Mythos 5 as a foundational tool for national and enterprise-level security operations. According to reporting by SOCFortress via Medium, a scan by Anthropic of 1,000 open-source projects using Mythos flagged over 23,000 potential vulnerabilities, with independent researchers confirming a 90% validity rate on high-severity findings. Cloudflare, one of the initial Glasswing partners, also reported that the model's false-positive rate was better than that of human testers, according to the same report.

The pricing of Fable 5 at $10 per million input tokens and $50 per million output tokens - roughly double the cost of Claude Opus 4.8 but less than half the price of the earlier Mythos Preview at $25 and $125 per million tokens respectively - positions the model as a premium but accessible offering for enterprise and developer customers. The Centre for Emerging Technology and Security at the Alan Turing Institute noted that Mythos Preview's pricing at five times that of Opus 4.6 and the commercial dimensions of Project Glasswing are difficult to fully disentangle from its stated security mission.

Latest Developments

The June 9, 2026 release of Claude Fable 5 and Mythos 5 follows a series of rapid developments since early 2026. Claude Mythos Preview was launched on April 7, 2026, through Project Glasswing, with approximately 50 founding partner organizations. On May 22, 2026, Anthropic published an initial update stating that it and its partners had used Mythos Preview to find more than 10,000 high- or critical-severity vulnerabilities across systemically important software worldwide. On June 2, 2026, Anthropic expanded Project Glasswing to roughly 150 new organizations across more than 15 countries.

The June 9 launch simultaneously made Fable 5 available on the Claude API, AWS Bedrock, Google Vertex AI, and Microsoft Foundry, and upgraded all current Glasswing partners from Mythos Preview to Mythos 5. According to the Amazon Bedrock model card, Fable 5 and Mythos 5 share a knowledge cutoff of December 2025 and support adaptive thinking via Anthropic's reasoning framework. Anthropic has indicated plans to expand access to Mythos 5 through a broader trusted access program in the future, though no timeline has been specified.

Top India News Analysis

The simultaneous release of a publicly available and a restricted version of the same underlying model reflects a calculated approach to managing capability risk at scale. By routing sensitive Fable 5 requests to Opus 4.8 rather than refusing them outright, Anthropic preserves developer experience while limiting misuse potential - a design choice that other frontier AI developers may study closely. However, independent technical scrutiny of Mythos-class cybersecurity claims, particularly the debate around orchestration versus raw model capability, suggests that the full picture of what these models can achieve autonomously versus with structured tooling remains an open question. The broader expansion of Project Glasswing to critical infrastructure sectors also raises governance questions that go beyond any single AI company's internal policy.

Key Takeaways

  • Anthropic released Claude Fable 5 and Claude Mythos 5 on June 9, 2026, with Fable 5 publicly available and Mythos 5 restricted to Project Glasswing partners.
  • Both models share the same underlying architecture; the distinction is in the application of safety safeguards.
  • Fable 5 is priced at $10 per million input tokens and $50 per million output tokens, representing a major price reduction from Mythos Preview.
  • The models lead benchmarks in software engineering, knowledge work, vision, and scientific research.
  • Mythos 5 has been used in drug design research, accelerating aspects of the process by approximately ten times.
  • Project Glasswing has now expanded to roughly 150 organizations across more than 15 countries, including critical infrastructure providers.
  • Fable 5 includes a fallback mechanism that routes certain sensitive queries to Claude Opus 4.8, triggering in fewer than 5% of sessions on average.
  • Independent researchers have raised questions about whether Mythos-class cybersecurity performance reflects frontier model capability or primarily effective orchestration infrastructure.
  • Anthropic has indicated plans to further expand Mythos 5 access through a broader trusted access program, with no confirmed timeline.

Sources Consulted

  • Anthropic Official Announcement - Claude Fable 5 and Claude Mythos 5 (anthropic.com/news/claude-fable-5-mythos-5), June 9, 2026
  • Anthropic Claude Mythos Product Page (anthropic.com/claude/mythos)
  • Amazon Bedrock Model Card - Claude Mythos Preview (docs.aws.amazon.com)
  • Anthropic Red Team - Mythos Preview Exploit Documentation (red.anthropic.com)
  • Centre for Emerging Technology and Security, The Alan Turing Institute - Claude Mythos: What Does Anthropic's New Model Mean for the Future of Cybersecurity? (cetas.turing.ac.uk)
  • CoderSera - Claude Fable 5 Launch Guide 2026 (codersera.com)
  • CoderSera - Anthropic Mythos: The Frontier Model Anthropic Withheld (codersera.com)
  • Let's Data Science - Anthropic launches Claude Fable 5 with guarded fallback (letsdatascience.com)
  • SOCFortress via Medium - Anthropic Expanding Project Glasswing and Claude Mythos Cyber Access (medium.com)
  • Gizmodo - Anthropic Lets Claude Mythos Spread Its Glasswings (gizmodo.com), June 2, 2026
  • flyingpenguin - Executive Summary for Claude Mythos Project Glasswing: June 2026 Verification Status (flyingpenguin.com)
  • Wikipedia - Claude Mythos (en.wikipedia.org)
  • Wikipedia - Claude Language Model (en.wikipedia.org)
  • claudefa.st - Claude Mythos Preview: Anthropic's Frontier Model (claudefa.st)

Author: Top India News Staff

Publisher: Top India News