The original headline is: "GPT-5.6 Sol matches Claude Fable 5 intelligence at one third the cost"

Short summary

OpenAI's GPT-5.6 family (Sol, Terra, Luna) launches with Sol matching Claude Fable 5's intelligence at roughly one-third the cost. Sol leads the Artificial Analysis Coding Agent Index at 80 and uses fewer output tokens than competitors. The family introduces cache-write pricing at 1.25× base input cost, mirroring Anthropic's model — a cost factor for agents with long shared contexts.

• •GPT-5.6 Sol matches Claude Fable 5 intelligence at ~1/3 the cost
• •Sol leads coding agent benchmarks at 80 on the AA Coding Agent Index
• •New cache-write pricing at 1.25× input cost impacts long-context agent budgets

Generated with AI, which can make mistakes.