Recently, Anthropic officially launched its next-generation flagship AI models Claude Fable 5.2 and Claude Mythos 5.2. The new versions focus on enhancing programming, logical reasoning, and visual understanding capabilities, with output speeds up to 30% faster than the previous generation, and a 20% reduction in API call prices. Benchmark tests released by Anthropic show that Fable 5.2 outperformed its own Opus 5.5 and the previous Fable 5.1 in multiple assessments.
In the FrontierCode 2.0 programming test, Fable 5.2 achieved a score of 49.1% at maximum reasoning intensity, surpassing Fable 5.1's 32.4% and Opus 5.5's 45.5%; in the FrontierSWE v2 test, it scored 69.9% (Fable 5.1 scored 56.3%, Opus 5.5 scored 62.3%); in the Humanity’s Last Exam test, it achieved a score of 74.2% when using tools; and in the RoboDojo Sim dual-arm robot simulation test, it scored 64.9% (the previous generation scored 27.6%).
In the Remote Labor Index test, it scored 41.3% (the previous generation scored 17.9%). However, in the AutomationBench enterprise workflow test, Fable 5.2 scored 51.4%, slightly lower than Google Gemini 4 Carbon's 53.9%.
Claude Fable 5.2 excels in programming tests
Anthropic introduced the practical performance of Fable 5.2 in internal software development. The company stated that over 90% of the merged code modifications in its core codebase were completed by Claude under human guidance. One key case involved using an early version of Fable 5.2 to migrate a critical production service from Python to Rust, resulting in approximately 26 times faster startup speed, along with improved operational efficiency and reliability. Additionally, the new model used fewer tokens to complete difficult tasks at moderate reasoning intensity, further reducing actual usage costs.
Anthropic's Fable 5.2 and Mythos 5.2 utilize the same underlying model but are configured with different limitations for different user groups. The publicly available Fable 5.2 has security mechanisms in place for cybersecurity and biology fields, with early tests showing a reduction of about 55% in security interceptions compared to Fable 5.1, allowing it to handle normal requests more flexibly. Mythos 5.2, on the other hand, is open to vetted professional organizations, primarily focusing on cybersecurity and life sciences research, and is not directly accessible to general users.
Pricing strategy for Fable 5.2 and Mythos 5.2
In terms of API pricing, both Fable 5.2 and Mythos 5.2 are priced at US$8 per million input tokens (approximately HK$62), and US$40 per million output tokens (the previous generation was US$10 (approximately HK$78) and US$50 (approximately HK$390)). Anthropic also offers a fast mode, with token generation speeds up to 2.5 times that of the normal mode, priced at US$16 per million input tokens (approximately HK$125) and US$80 per million output tokens (approximately HK$624). Currently, Fable 5.2 is available to paid users of Claude and is offered via API, supporting platforms such as AWS, Google Cloud, and Microsoft Azure.
Eligible paid subscribers will receive 1 to 3 additional usage quota resets this week.

