OpenAI Releases GPT-5.6 (Sol, Terra, Luna): A Three-Tier Model Family With Programmatic Tool Calling in the Responses API

Short summary

OpenAI released GPT-5.6 in three tiers (Sol, Terra, Luna) with Programmatic Tool Calling—a feature that runs model-generated JavaScript in isolated V8 to orchestrate tools directly, reducing prompt token usage by 38-63% versus traditional approaches. Sol achieves 80 on the Artificial Analysis Coding Agent Index and 62.6% on OSWorld 2.0 benchmarks. While Claude Fable 5 leads in overall intelligence metrics, GPT-5.6's tool automation architecture significantly improves efficiency for agent-based applications.

• •GPT-5.6 ships in three tiers (Sol/Terra/Luna) with performance-cost tradeoffs; Sol costs $5-30/1M tokens
• •Programmatic Tool Calling runs model-generated JavaScript in isolated runtime, reducing token usage by 38-63%
• •Sol matches Claude on coding benchmarks but Fable 5 still leads on overall intelligence metrics

Generated with AI, which can make mistakes.