OpenAI Unveils GPT-5.6 Model With 1.5 Million Token Context Window

OpenAI is preparing to release GPT-5.6, a new iteration of its flagship language model series that significantly expands the context window available for processing long documents and complex codebases.

New model expands context window to handle long documents and complex codebases

The upcoming model operates under the internal development codename iris-alpha, which OpenAI has identified in backend system logs alongside other potential version identifiers such as ember-alpha and beacon-alpha.

GPT-5.6 supports a maximum context window of 1.5 million tokens, representing a substantial increase from the current GPT-5.5 API limit of 1.05 million tokens.

Early real-world testing by developers using OpenCode suggests that the model maintains stable performance when processing inputs up to 900,000 tokens and handles requests exceeding the 1.05 million token threshold without degradation.

OpenAI expects GPT-5.6 to launch in June 2026, positioning it against competing models such as Anthropic's Claude Sonnet 4.8 and Google's Gemini 3.5 Pro that are targeting a similar release window.