GPT-5.6 Sol vs. Claude Fable 5: 7 Things to Know

On July 9, OpenAI rolled out GPT-5.6 Sol globally and the model’s pricing came in well below its biggest rival, Anthropic.

Both companies charge by the token, the small pieces of text AI models read and generate. Sol charges $5 for every million tokens you send it and $30 for every million it writes back. Claude Fable 5 by Anthropic charges double, $10 and $50 for the same.

The price difference is notable on its own. But what makes GPT-5.6 Sol vs Fable 5 more interesting is that Sol outperforms Fable 5 on two major tests the companies have published results for.

On Agents' Last Exam, a test of how well a model handles long, real-world jobs on its own, Sol scored 13.1 points higher than Fable 5. On Terminal-Bench 2.1, which measures how well a model can code and complete tasks with minimal help, Sol scored 88.8% versus Fable 5's 83.4%.