Anthropic has launched Claude Fable 5.1, a new artificial intelligence model designed for coding, research, knowledge work and long-running autonomous tasks. The company also introduced Claude Mythos 5.1, a version with more permissive safeguards available only to approved cybersecurity and life-sciences organizations.
Fable 5.1 is available immediately across Claude’s platforms and major cloud providers. Anthropic says the model offers stronger performance than Fable 5 while reducing the cost of typical token-billed workloads by approximately 25%.
Claude Fable 5.1: Key details
- Available now across Claude platforms
- API model name: claude-fable-5-1
- About 25% cheaper for typical workloads
- Potential savings of around 45% for highly agentic tasks
- $10 per million input tokens
- $50 per million output tokens
- Cache reads reduced to $0.25 per million tokens
- Mythos 5.1 access is restricted to vetted organizations
What is new in Claude Fable 5.1?
Anthropic describes Fable 5.1 as its most advanced generally available model for coding and professional knowledge work. The model is intended to handle complex assignments that require planning, tool use, repeated verification and sustained work over longer periods.
The company says Fable 5.1 is less likely to rely on shortcuts that produce incomplete results. It has instead been trained to investigate underlying problems, verify its work and maintain direction during complicated, multi-step tasks.
That could make the update particularly relevant to software developers using Claude Code for debugging, code review, application development and incident investigation.
In one test cited by Anthropic, Fable 5.1 reportedly identified the cause of a rare software crash that engineers at investment firm Millennium had been unable to explain for several years. The account was supplied by an early-access customer and should be understood as a company-presented example rather than an independently controlled benchmark.
Anthropic also published several benchmark results comparing the model with Fable 5, Opus 5 and OpenAI’s GPT-5.6 Sol. Fable 5.1 recorded a score of 55.8% on Terminal-Bench 4.0, compared with 42% for Fable 5, 52.3% for Opus 5 and 37.3% for GPT-5.6 Sol.
On CursorBench 3.2.0, Anthropic reported a score of 73.4% for Fable 5.1, ahead of Fable 5 at 70.5%, Opus 5 at 70% and GPT-5.6 Sol at 67.2%.
These results were published by Anthropic using its own evaluation setup. Benchmark scores can vary according to testing conditions, effort settings, tools and safeguards, meaning they should not be treated as proof that one model will perform better in every real-world situation.
The launch arrives as competition among frontier AI developers continues to intensify. OpenAI, Anthropic and other technology companies are increasingly focusing on AI agents that can perform complete workflows instead of responding only to individual prompts. The regulatory environment is also changing, as detailed in our report on the tougher EU rules affecting ChatGPT and AI platforms.
Claude Fable 5.1 price, access and Mythos restrictions
Fable 5.1 retains the headline price of Fable 5 at $10 per million input tokens and $50 per million output tokens. The largest pricing change applies to cache reads, which now cost $0.25 per million tokens—a reduction of 75%.
Cache reads allow the model to reuse previously processed context instead of processing the same information again. This can produce substantial savings for coding agents and other applications that repeatedly work with large codebases, documents or tool results.
Anthropic estimates that the change will lower the cost of a typical workload by approximately 25% compared with Fable 5. For context-heavy and tool-heavy agentic work, estimated savings could reach around 45%. Actual costs will depend on usage patterns and the amount of cached context reused.
| Usage | Fable 5.1 price |
|---|---|
| Input tokens | $10 per million |
| Output tokens | $50 per million |
| Cache reads | $0.25 per million |
Claude Fable 5.1 is available through Claude, Claude Code and the Claude API. Anthropic says the model is also being offered through Amazon Web Services, Google Cloud and Microsoft Azure.
Mythos 5.1 uses the same underlying model as Fable 5.1 but operates with more permissive safeguards for specialist work. Its availability is currently limited to vetted cybersecurity professionals and life-sciences researchers through Anthropic’s trusted-access programs.
According to Anthropic, Mythos 5.1 is the strongest cybersecurity model it has released. Because those capabilities could potentially be misused, the company is restricting access while working with the US government to expand the program to more domestic and international partners.
Fable 5.1 also introduces more precise cybersecurity controls. Anthropic says Claude Code users should encounter approximately 60% fewer safeguard interventions during legitimate security work. The generally available model can help identify software vulnerabilities, but tasks such as exploit generation and certain forms of penetration testing remain restricted or redirected.
The company has also announced Enterprise Frontier Safeguards, a system intended to let enterprise customers retain their data inside customer-controlled cloud infrastructure. Anthropic expects the system to begin rolling out in phases later this fall.
Full technical results, safety disclosures and access information are available in Anthropic’s official Fable 5.1 and Mythos 5.1 announcement.
Disclosure: Benchmark results and cost-saving estimates in this report were supplied by Anthropic. Performance and actual API costs may differ according to workload, configuration and usage.
Make Swikblog your go-to source on Google for reliable updates, smart insights, and daily trends.
