GPT-5.6 model releases

OpenAI's cybersecurity-specialized model, launched Aug 10 2026 as the successor to the limited-preview GPT-5.5-Cyber. Built on GPT-5.6-Sol and tuned for authorized security workflows — vulnerability discovery, exploit-chain development, privilege escalation, authentication-bypass analysis, patch validation, and red-teaming — with a deliberately lower refusal rate on dual-use security tasks than general GPT-5.6. OpenAI reports a 95% completion rate on advanced exploit-chain prompts (vs ~1.5% for GPT-5.6-Sol) and that the model surfaced a high-severity V8 JavaScript-engine vulnerability plus flaws in unnamed mobile-OS, database, and kernel targets. Not generally available: access is gated to vetted defenders through the expanded Daybreak program's new "Daybreak Red" tier (partners include Accenture, IBM, Palo Alto Networks, CrowdStrike, Fortinet, Akamai, and Cloudflare). Size undisclosed; weights not released.

Balanced, everyday-work tier of OpenAI's GPT-5.6 series, officially previewed June 26, 2026. OpenAI positions Terra as competitive with GPT-5.5 while being roughly 2x cheaper. Shares the series' new `max` reasoning effort and `ultra` subagent mode and OpenAI's GPT-5.6 safety stack. Begins as a limited preview via the API and Codex for vetted partners after U.S. government review, with broad availability planned in the following weeks. Priced at $2.50 / $15 per million input/output tokens.

Flagship tier of OpenAI's GPT-5.6 series, officially previewed June 26, 2026 — OpenAI's strongest model to date. Adds a new `max` reasoning effort for the deepest reasoning and an `ultra` mode that uses subagents to accelerate complex work. Sets a new state of the art on Terminal-Bench 2.1 (command-line, agentic coding) and shows broad gains in long-horizon biology (GeneBench v1) and cybersecurity (ExploitBench, ExploitGym), paired with OpenAI's most robust safety stack and a phased release. Begins as a limited preview via the API and Codex for a small group of vetted partners after U.S. government review, with general availability planned in the following weeks; also launching on Cerebras at up to 750 tokens/sec in July. Priced at $5 / $30 per million input/output tokens.

Fast, low-cost tier of OpenAI's GPT-5.6 series, officially previewed June 26, 2026 — the most affordable model in the family, bringing strong capability at OpenAI's lowest cost. Shares the series' capability-tier naming and GPT-5.6 safety stack. Begins as a limited preview via the API and Codex for vetted partners after U.S. government review, with broader availability planned in the following weeks. Priced at $1 / $6 per million input/output tokens.