Positioning
GPT-5.5 is OpenAI’s flagship reasoning model for complex professional work, including coding and multi-step tasks involving tools. The API alias is gpt-5.5; its documented default snapshot is gpt-5.5-2026-04-23. (openai.com)
Capabilities and uses
- Writes and debugs code, researches online, analyzes data, creates documents and spreadsheets, and helps with scientific research and other knowledge work.
- Accepts text and image input and produces text output. It supports function calling, structured outputs and, through the Responses API, tools including web search, code interpreter and computer use.
- The documented context window is 1,050,000 tokens, with up to 128,000 output tokens. Reasoning effort can be set to
none,low,medium,highorxhigh; the default ismedium. (openai.com)
Limitations
The model’s documented knowledge cutoff is December 1, 2025. Audio and video input are not supported, and fine-tuning is not supported. Image generation is available through a Responses API tool rather than as native image output from this model. Results and tool actions should be checked before reliance, particularly for consequential work. GPT-5.5 prompt caching requires extended prompt caching; in-memory caching is not supported. (developers.openai.com)
Release
OpenAI announced GPT-5.5 on April 23, 2026. Its announcement was updated on April 24, 2026 to state that GPT-5.5 was available in the API. (openai.com)
API pricing
OpenAI API text-token prices, in USD per 1 million tokens, checked November 16, 2026. These are for gpt-5.5, not GPT-5.5 Pro. (developers.openai.com)
| Processing | Input tokens | Input | Cached input | Output |
|---|---|---|---|---|
| Standard | ≤272,000 | $5.00 | $0.50 | $30.00 |
| Standard | >272,000 | $10.00 | $1.00 | $45.00 |
| Batch | ≤272,000 | $2.50 | $0.25 | $15.00 |
| Batch | >272,000 | $5.00 | $0.50 | $22.50 |
| Flex | ≤272,000 | $2.50 | $0.25 | $15.00 |
| Flex | >272,000 | $5.00 | $0.50 | $22.50 |
| Fast | ≤272,000 | $12.50 | $1.25 | $75.00 |
For prompts exceeding 272,000 input tokens, the long-context rates apply to the full session for Standard, Batch and Flex. Fast-mode long-context pricing was not verified. Regional processing endpoints for data residency carry a 10% uplift. Tool calls can incur separate charges. (developers.openai.com)