Home / Technology / New AI Fable Beats GPT-5.5 and Gemini in Tests
New AI Fable Beats GPT-5.5 and Gemini in Tests
10 Jun
Summary
- Fable 5 outperforms previous models and rivals in complex tasks.
- New AI excels in software engineering, drug design, and vision.
- Subscribers can test Fable 5 until June 22 without charge.
Anthropic has launched Fable 5, a new family of AI models building on its state-of-the-art Mythos system. Fable 5 offers capabilities exceeding previous public releases, particularly excelling in longer and more complex tasks. In internal testing, Fable 5 outperformed Anthropic's Opus 4.8, as well as OpenAI's GPT-5.5 and Google's Gemini 3.1 Pro.
The new model shows particular strength in software engineering, drug design, and document analysis. It also significantly improves upon previous models in vision tasks, capable of processing detailed scientific figures and reconstructing web application code from screenshots. This enhanced vision capability allows Fable 5 to successfully play games like FireRed with minimal input.
Anthropic has implemented safeguards in Fable 5, which may occasionally flag harmless requests, occurring in less than 5% of sessions. The company also released Mythos 5, based on the same model but with fewer safeguards, initially for select partners. Claude subscribers can access Fable 5 until June 22 without charge, after which it will be available via usage credits.
