Changes from previous version
First Sol-tier release under the durable Sol/Terra/Luna naming system introduced with GPT 5.6. Step up from GPT 5.5 on Terminal-Bench 2.1, GeneBench v1 token efficiency, and long-horizon cyber benchmarks; adds max and ultra agentic modes plus layered safeguards not present on the legacy GPT product line.
Release Summary
OpenAI's strongest model yet and flagship of the new GPT 5.6 tier naming: the generation number (5.6) advances on its own cadence while Sol, Terra, and Luna are durable capability tiers. State-of-the-art on Terminal-Bench 2.1 for agentic command-line coding; stronger GeneBench v1 genomics results than GPT 5.5 with fewer tokens; on ExploitBench² competitive with Claude Mythos Preview using roughly one-third of the output tokens. Introduces max reasoning effort and ultra mode (parallel subagents). Ships with OpenAI's most robust safety stack to date. Limited preview via API and Codex for trusted partners; broader ChatGPT and API availability planned in coming weeks. API pricing $5/M input and $30/M output, matching the prior GPT 5.5 flagship. Positioned against Claude Fable 5 and Claude Mythos 5 on frontier agentic and cyber workloads.
Timeline
GPT-5.6 Sol preview begins
OpenAI starts a limited preview of GPT-5.6 Sol alongside Terra and Luna through the API and Codex for trusted partners, with expanded availability planned in coming weeks.
GPT-5.6 tier naming introduced
OpenAI introduces durable Sol, Terra, and Luna tiers that can advance on their own cadence while the generation number identifies the model family (5.6).