OpenAI Releases GPT-5.6 Family
Simon Willison · rss · 2026-07-10
OpenAI has released the new generation GPT-5.6 family: three tiers named Luna, Terra, and Sol, which are now generally available.
Pricing and Specs
- Luna: $1 / million input tokens, $6 / million output tokens
- Terra: $2.5 / million input tokens, $15 / million output tokens
- Sol: $5 / million input tokens, $30 / million output tokens
- All three models support a 1 million token context, a maximum output of 128,000 tokens, and a knowledge cutoff of 2026-02-16.
OpenAI's Core Selling Points
OpenAI emphasizes that GPT-5.6 is stronger in long-running agent workflows. Cited evaluations show that in long-term professional task benchmarks like Agents’ Last Exam (covering 55 domains), Sol scored 53.6, surpassing Claude Fable 5; Terra and Luna also beat Fable 5 at a lower cost.
New API Capabilities
The article and evaluations also mentioned several noteworthy new features:
- Programmatic Tool Calling: Models can orchestrate and run JavaScript to organize tool calls.
- Multi-agent: Can directly spawn parallel sub-agents within the core API.
- Prompt cache breakpoints: Supports explicitly setting cache breakpoints.
- Image requests can now set detail: original to avoid preprocessing scaling.
The author also noted that OpenAI's own audit of SWE-Bench Pro found that about 30% of the tasks in the benchmark might be problematic.
More from coding & agent
- A roundup of AI agents and MCP resources, including how to evaluate agents — _jaydeepkarale · 2026-07-21
- A full course shows how to build and deploy an AI agent with OpenAI and LangChain — _jaydeepkarale · 2026-07-21
- A beginner guide to AI agents points readers to a Stanford webinar — _jaydeepkarale · 2026-07-21
- A practical guide on how to evaluate AI agents — _jaydeepkarale · 2026-07-21
- MCP is headed toward easier scale, event-driven extensions, and workable file uploads — EricBuess · 2026-07-21
- Developers debate the missing composition model for AI agents — threepointone · 2026-07-21