Overview
- OpenAI has listed GPT-5.6 Sol at $5 per million input tokens and $30 per million output tokens, versus Claude Fable 5 at $10 and $50 respectively, producing a roughly 42 percent blended saving in a 1M in + 1M out example.
- Vendor-cited benchmarks reported higher scores for Sol on agentic and coding tests, including a Terminal-Bench 2.1 result of 88.8 percent for Sol versus 84.3 percent for Fable 5.
- OpenAI CEO Sam Altman amplified the cost and efficiency claims in a post on X on Tuesday, saying Sol is "half the price" and about "twice as token efficient" in many cases.
- Reporting highlights Sol’s strengths for execution, command-line and API-heavy coding pipelines while noting Fable 5 is portrayed as relatively stronger on deep analysis and creative tasks.
- Independent verification and an on-record response from Anthropic are not present in the coverage and buyers should watch for third‑party benchmarks and real‑world tests before changing deployments.