← All stories
● Covered by 9 sources · 14 reportsMedium impact13 neutral1 positive

Anthropic Releases Claude Fable 5.2 and Mythos 5.2 with Speed, Cost, and Capability Upgrades

🔄 Updated 5d ago — new reporting from BleepingComputer
New to BrevFeed? We gather this story from every outlet covering it into one summary — ranked by real-world impact, not just the latest headline — so you never miss what matters. What is BrevFeed? →

Key points

  • Claude Fable 5.2 and Mythos 5.2 are Anthropic's latest models for technical challenges.
  • Models generate outputs up to 30% faster than previous 5.1 versions.
  • Input token pricing for Fable 5.2 is reduced to $8 per million.
  • Improvements noted in coding, reasoning, and vision capabilities.
  • Fable 5.2 is generally available; Mythos 5.2 requires verification for access.
  • Fable 5.2 achieved a 76.3% completion rate on Irregular's CyScenarioBench.
  • Fable 5.2 surpassed GPT-6 Cyber's 61% on CyScenarioBench.
  • Models showed progress in vulnerability rediscovery and cybersecurity engineering.
  • Models helped engineers complete software migrations and design electronic devices.
  • Models solved two open problems in black hole physics and formalized Fermat’s Last Theorem.
  • Fable 5.2 was an order of magnitude less likely to publish harmful content than Opus 5.5 and Sonnet 5.5.
  • Mythos 5.2 roughly doubled Mythos 5.1's vulnerability discovery results in Firefox testing.
  • Mythos 5.2 achieved an 89.6% success rate in CVE rediscovery, compared to 78.3% for Mythos 5.1.
  • Mythos 5.2 showed reduced sandbox escapes.
  • Fable 5.2 has protections against cybersecurity and biology misuse.
  • Mythos 5.2 ships without some of Fable 5.2's guardrails.
  • Fable 5.2 generated a playable browser-based first-person shooter game named "Saltline."
  • Fable 5.2 created "Saltline" from one prompt over eight hours, costing an estimated $550.
  • Fable 5.2 processed 445 million input tokens and produced 2.7 million output tokens for "Saltline."
  • Anthropic recommends Opus 5.5 for most daily coding and knowledge work.
  • Fable 5.2 scored 45.3% on FrontierCode 2.0 at High effort.
  • Claude Fable 5.2 launched on November 9.
  • Fable 5.2 scored 49.1% on FrontierCode 2.0 at X high effort.
  • Fable 5.2 scored 69.9% on FrontierSWE v2, surpassing GPT-6 Astra's 65.5%.
  • GPT-6 Astra launched in September.
  • GPT-6 Astra's standard API rates are $10 per million input tokens and $50 per million output tokens.
  • Fable 5.2 output token pricing is $40 per million.
  • Fable 5.2 triggered 55% fewer safeguard blocks than Fable 5.1 overall.
  • Fable 5.2 triggered 60% fewer safeguard blocks than Fable 5.1 in cybersecurity.
  • Fable 5.1's standard API rates were $10 per million input tokens and $50 per million output tokens.
  • Opus 5.5 scored 45.5% on FrontierCode 2.0 at Max effort.
  • Opus 5.5 scored 62.3% on FrontierSWE v2.
  • Opus 5.5 input token pricing is $4 per million.
  • Opus 5.5 output token pricing is $20 per million.
  • Mythos 5.2 is Anthropic's most capable model for cybersecurity engineering.
  • Mythos 5.2 detects 10.2% more vulnerabilities than Opus 5.5.
  • Mythos 5.2 detects 14.4% more vulnerabilities than Mythos 5.1.
  • One in six bugs found by Mythos 5.2 were missed by other frontier models.

New Models for Technical Challenges

Anthropic has launched Claude Fable 5.2 and Claude Mythos 5.2, its newest AI models. These models are designed to address demanding technical challenges and are considered Anthropic's most capable models for such tasks. Claude Fable 5.2 is available on platforms including Amazon Bedrock, Google Cloud, and Microsoft Azure, while Claude Mythos 5.2 is accessible to organizations through Anthropic’s Life Sciences and Cyber Verification Programs.

Performance and Cost Improvements

The updated models generate outputs up to 30% faster than their predecessors, Claude Fable 5.1 and Claude Mythos 5.1. Anthropic has also reduced the pricing for Fable 5.2 to $8 per million input tokens and $40 per million output tokens. An optional 'Fast mode' is available for even quicker responses, though it comes at double the standard price.

Enhanced Capabilities

Anthropic states that Fable 5.2 and Mythos 5.2 show improvements in coding, reasoning, and vision. During prerelease testing, the models assisted engineers with software migrations, designed electronic devices, solved theoretical physics problems, and formalized Fermat’s Last Theorem. They are also reported to communicate more clearly and exhibit reduced hallucination rates compared to previous versions.

Model Distinctions and Access

Claude Fable 5.2 and Claude Mythos 5.2 share the same underlying model. However, Fable 5.2 includes stronger cybersecurity and biology safeguards, making it generally available. Mythos 5.2 is intended for specialized beneficial applications in cybersecurity and biomedical research, requiring access through specific verification programs. Fable 5.2 also supports zero data retention (ZDR) policies via the Enterprise Frontier Safeguards system.

Positioning in the Claude Family

Within Anthropic's model hierarchy, Fable is positioned as the most capable model, above Haiku, Sonnet, and Opus. While Opus 5.5 is recommended for most coding and knowledge tasks, Fable 5.2 is specifically geared towards the most ambitious and open-ended technical work.

Updates

🕒 2026-11-16 · new reporting from BleepingComputer
  • Mythos 5.2 is Anthropic's most capable model for cybersecurity engineering.
  • Mythos 5.2 detects 10.2% more vulnerabilities than Opus 5.5.
  • Mythos 5.2 detects 14.4% more vulnerabilities than Mythos 5.1.
  • One in six bugs found by Mythos 5.2 were missed by other frontier models.
🕒 2026-11-15 · new reporting from The New Stack
  • Opus 5.5 scored 45.5% on FrontierCode 2.0 at Max effort.
  • Opus 5.5 scored 62.3% on FrontierSWE v2.
  • Opus 5.5 input token pricing is $4 per million.
  • Opus 5.5 output token pricing is $20 per million.
🕒 2026-11-14 · new reporting from The New Stack
  • Claude Fable 5.2 launched on November 9.
  • Fable 5.2 scored 49.1% on FrontierCode 2.0 at X high effort.
  • Fable 5.2 scored 69.9% on FrontierSWE v2, surpassing GPT-6 Astra's 65.5%.
  • GPT-6 Astra launched in September.
  • GPT-6 Astra's standard API rates are $10 per million input tokens and $50 per million output tokens.
  • Fable 5.2 output token pricing is $40 per million.
  • Fable 5.2 triggered 55% fewer safeguard blocks than Fable 5.1 overall.
  • Fable 5.2 triggered 60% fewer safeguard blocks than Fable 5.1 in cybersecurity.
  • Fable 5.1's standard API rates were $10 per million input tokens and $50 per million output tokens.
🕒 2026-11-11 · new reporting from The New Stack
  • Fable 5.2 generated a playable browser-based first-person shooter game named "Saltline."
  • Fable 5.2 created "Saltline" from one prompt over eight hours, costing an estimated $550.
  • Fable 5.2 processed 445 million input tokens and produced 2.7 million output tokens for "Saltline."
  • Anthropic recommends Opus 5.5 for most daily coding and knowledge work.
  • Fable 5.2 scored 45.3% on FrontierCode 2.0 at High effort.
🕒 2026-11-10 · new reporting from SecurityWeek
  • Mythos 5.2 roughly doubled Mythos 5.1's vulnerability discovery results in Firefox testing.
  • Mythos 5.2 achieved an 89.6% success rate in CVE rediscovery, compared to 78.3% for Mythos 5.1.
  • Mythos 5.2 showed reduced sandbox escapes.
  • Fable 5.2 has protections against cybersecurity and biology misuse.
  • Mythos 5.2 ships without some of Fable 5.2's guardrails.
🕒 2026-11-10 · new reporting from The Hacker News
  • Fable 5.2 achieved a 76.3% completion rate on Irregular's CyScenarioBench.
  • Fable 5.2 surpassed GPT-6 Cyber's 61% on CyScenarioBench.
  • Models showed progress in vulnerability rediscovery and cybersecurity engineering.
  • Models helped engineers complete software migrations and design electronic devices.
  • Models solved two open problems in black hole physics and formalized Fermat’s Last Theorem.
  • Fable 5.2 was an order of magnitude less likely to publish harmful content than Opus 5.5 and Sonnet 5.5.

✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →

The daily brief

One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.

One email a day. Unsubscribe in one click, any time.

Today's brief

Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.

~4 min · 3 stories · Nov 21

▶ Play today's brief Listen on Spotify

New every morning, and the back catalogue is archived by date.

How outlets covered it

Anthropic released Claude Mythos 5.2 to partners and Fable 5.2 to general users, with Mythos 5.2 showing improved capabilities in finding and fixing software vulnerabilities. Fable 5.2 retains security safeguards but has reduced unnecessary blocking of cybersecurity-related requests by approximately 60% compared to its predecessor.

Anthropic's Claude Fable 5.2 and Opus 5.5 models achieved similar scores in independent coding and agent tests, despite Fable 5.2 having higher API rates. Fable 5.2's standard API rates are double those of Opus 5.5, costing $8 per million input tokens and $40 per million output tokens, compared to Opus's $4 and $20.

Anthropic's Claude Fable 5.2 demonstrates fewer safeguard blocks and reduced API pricing compared to Fable 5.1. Independent testing on developer tasks confirmed fewer refusals and some cost savings, though not as pronounced as Anthropic's claims in all scenarios.

Anthropic's Claude Fable 5.2, launched November 9, shows higher benchmark scores and lower per-token API pricing than OpenAI's GPT-6 Astra. However, practical testing on developer tasks reveals that Fable 5.2 often incurs higher overall costs due to greater token consumption, despite its lower listed rates.

Anthropic released Fable 5.2, a new coding model that surpasses Opus 5.5 on two benchmarks and can complete certain tasks at half the cost. Despite Fable's performance, Anthropic recommends Opus 5.5 for most daily coding and knowledge work due to its overall capabilities and pricing.

Anthropic's Claude Fable 5.2 generated a playable browser-based first-person shooter game, "Saltline," from one prompt over eight hours, costing an estimated $550 in API usage. The model processed 445 million input tokens and produced 2.7 million output tokens to create the game, which features procedurally generated visuals.

Anthropic released Claude Mythos 5.2 and Fable 5.2, with Mythos 5.2 demonstrating improved vulnerability discovery, roughly doubling the results of Mythos 5.1 in Firefox testing. Mythos 5.2 achieved an 89.6% success rate in CVE rediscovery, compared to 78.3% for Mythos 5.1, and also showed reduced sandbox escapes.

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, demonstrating improvements in coding, reasoning, and vision. Fable 5.2 achieved a 76.3% completion rate on Irregular's CyScenarioBench, surpassing GPT-6 Cyber's 61%. The models also showed progress in vulnerability rediscovery and cybersecurity engineering capabilities.

Anthropic has made Claude Fable 5.2 and Claude Mythos 5.2 available on Amazon Bedrock and the Claude Platform on AWS. Claude Fable 5.2 is designed for technical work, including coding, math, and science, while Claude Mythos 5.2 is for advanced work in cybersecurity and life sciences, accessible through verification programs.

Anthropic released new versions of its Mythos and Fable models, improving coding, reasoning, and vision capabilities. Fable 5.2's cost was reduced to $8 per million input tokens, making it cheaper than GPT-6 Astra, and it offers 30% faster response times. The release follows a report suggesting Anthropic's internal models are accelerating its AI progress, raising questions about recursive self-improvement.

Anthropic launched Fable 5.2 and Mythos 5.2, updating its flagship models with up to 30% faster output generation and a 20% price reduction for standard use. These models also introduce a 'Fast mode' for even quicker responses and show improved coding scores, positioning them for complex technical challenges.

Anthropic has released Claude Fable 5.2 and Mythos 5.2, its latest AI models, which offer up to 30% faster output generation and reduced pricing. These models show improvements in coding, reasoning, and vision, and are intended for complex technical work.

Anthropic has launched Claude Fable 5.2 and Claude Mythos 5.2, new AI models with claimed improvements in coding, reasoning, and vision, and faster output generation. Claude Fable 5.2 is available immediately, while Claude Mythos 5.2 is accessible to vetted organizations, with both models intended for complex technical challenges.

Anthropic has released Claude Fable 5.2 and Claude Mythos 5.2, updated AI models with improvements in coding, reasoning, and vision capabilities. These models offer faster output generation, reduced pricing, and an optional fast mode, making them suitable for complex technical challenges.