OpenAI Launches GPT-5.5 and GPT-5.5 Pro in API

April 24, 2026

Details

  • OpenAI released GPT-5.5 and GPT-5.5 Pro models in the API on April 24, 2026, as announced by CEO Sam Altman.
  • GPT-5.5 excels in complex production workflows like coding, tool-heavy agents, grounded assistants, long-context retrieval, and customer-facing tasks requiring high execution quality.
  • Key upgrades include a native 1M token context window for handling massive codebases or documents, improved image inputs preserving more visual detail up to 10M pixels, and better instruction following.
  • Supports all prior API features like prompt caching, hosted tools, and new reasoning.effort tuning (low to xhigh) for latency/performance balance.
  • GPT-5.5 Pro offers enhanced capabilities, rolling out to ChatGPT Plus/Pro/Business/Enterprise and Codex users; priced higher at $5/$30 per million input/output tokens.
  • Benchmarks show strengths in agent tasks (OSWorld 78.7% vs. human 72.4%), token efficiency (40% fewer tokens vs. GPT-5.4), with matching latency.

Impact

OpenAI's GPT-5.5 API release intensifies competition with Anthropic's Claude and Google's Gemini, advancing agentic AI for autonomous multi-step tasks like code resolution and data analysis. The 1M token context and structured JSON outputs reduce developer friction, potentially accelerating enterprise adoption in coding and workflows. Higher pricing reflects premium capabilities, but token efficiency improvements lower effective costs compared to GPT-5.4. Enhanced safeguards align with regulatory pressures on AI safety, positioning OpenAI to lead in production-grade frontier models amid rising demands for reliable, tool-integrated intelligence.

Read more