Anthropic's Claude Fable 5 tops benchmarks; state-of-the-art on 95% of tested tasks

Anthropic released Claude Fable 5, a new frontier model that achieves state-of-the-art performance on nearly all tested benchmarks, including coding, reasoning, and multimodal tasks. The model combines Anthropic's constitutional AI approach with training improvements that reduce hallucination and improve few-shot learning across enterprise use cases.

Fable 5 targets knowledge-worker automation and agentic reasoning; Anthropic positioned the release as a direct competitor to OpenAI's o1 and Google's Gemini 2.0 Ultra. Benchmark deltas suggest meaningful improvements in long-context retrieval and code generation, signaling closer parity with closed-weight giants on real-world productivity workflows.