Phoenix 24
Higher token prices accompany stronger reasoning, broader tool use and greater autonomy.
SAN FRANCISCO, CALIFORNIA
OpenAI has introduced GPT-6 Astra, describing it as its most capable model for complex reasoning, programming, scientific research, computer use and professional work. The model is designed to complete multistep workflows across applications, browsers and documents while exercising stronger judgement over task boundaries.
Astra supports five reasoning levels, ranging from low to max, but does not offer a reasoning-free mode. It can use web and file search, computer control, code execution, image generation and external tools. New capabilities include asynchronous tool calls, instructions submitted while a response is underway and changes to reasoning intensity during an active conversation.
The model has a context window of 1.05 million tokens and can generate up to 128,000 output tokens. Its knowledge cutoff is April 30, 2026. It accepts text and images as input but does not natively support audio or video. OpenAI says Astra can achieve stronger results with fewer output tokens in some evaluations, potentially reducing the total cost of completing complex tasks despite its higher price per token.
Standard API pricing is $10 per million input tokens, $1 for cached input, $12.50 for cache writes and $50 per million output tokens. Prompts exceeding 272,000 input tokens cost twice the standard input and cache rates, while output is charged at 1.5 times the usual rate. Batch and Flex processing receive a 50 percent discount.
GPT-6 Astra is being introduced across ChatGPT Work and Codex, but availability depends on the user’s plan, account, rollout status and workspace configuration. Enterprise organisations require administrator activation during the initial rollout. API access is managed separately and is not automatically included with a ChatGPT subscription.
Detrás de cada dato, la intención. / Behind every data point, the intention.