OpenAI launched GPT-5.6, a family of three models — Sol (flagship), Terra (balanced), and Luna (cost-efficient) — that set new state-of-the-art results across coding, knowledge work, cybersecurity, and science while using fewer tokens at lower cost than competitors. On August 30, OpenAI cut Luna pricing by 80% and Terra by 20%.
Ultra mode: parallel multi-agent
The standout feature is ultra, OpenAI’s highest-capability setting that coordinates four agents in parallel by default. Ultra trades higher token consumption for stronger results and faster time-to-result on demanding tasks. On BrowseComp and SEC-Bench Pro, 16-agent configurations push the score-latency frontier further. Developers can build similar experiences using the multi-agent beta in the Responses API.
Coding state of the art
GPT-5.6 Sol with max reasoning scores 80 on the Artificial Analysis Coding Agent Index — 2.8 points above Claude Fable 5 — while using less than half the output tokens, taking less than half the time, and costing about one-third less. Terra performs just above Fable 5; Luna outperforms Claude Opus 4.8. The family also sets new records on Terminal-Bench 2.1 and DeepSWE for complex CLI workflows and real-codebase engineering.
Programmatic Tool Calling
GPT-5.6 writes and runs lightweight programs that coordinate tools, filter intermediate data, and adapt workflows dynamically. Programmatic Tool Calling in the Responses API eliminates the need to pass every tool response back through the model, reducing token consumption and round trips for tool-heavy tasks.
Knowledge work and design
Sol achieves 92.2% on BrowseComp and 62.6% on OSWorld 2.0 (surpassing Opus 4.8 with 85% fewer output tokens). The model creates fully editable presentations from scratch, infers design systems from reference decks, and produces refined documents and spreadsheets with improved typography, hierarchy, and layout precision.