product updateOpenAI

OpenAI lists GPT-6.2 Astra Pro: pro reasoning mode of GPT-6.2 Astra, $8/$40 per 1M tokens, 1.1M context

TL;DR

OpenAI's GPT-6.2 Astra Pro is the same underlying model as GPT-6.2 Astra, served with reasoning.mode set to pro. OpenRouter lists it at $8 input and $40 output per 1M tokens with a 1.1M context window, released Nov 17, 2026.

1 min read
0

OpenAI's GPT-6.2 Astra Pro is the same underlying model as GPT-6.2 Astra, served with reasoning.mode set to pro for higher-quality responses on complex tasks, according to its OpenRouter listing (openai/gpt-6.2-astra-pro). The listing gives a release date of Nov 17, 2026.

Key specifications

  • Model ID: openai/gpt-6.2-astra-pro
  • Underlying model: GPT-6.2 Astra (same model, different reasoning mode)
  • Context window: 1.1M tokens
  • Pricing: $8 per 1M input tokens, $40 per 1M output tokens
  • Released: Nov 17, 2026
  • EU availability: Not currently available through EU in-region routing

Parameter count, benchmark scores and training cutoff date are not stated in the listing.

Cost and latency

The listing says pro mode spends far more reasoning tokens per request. A typical request therefore costs several times more than the same request on GPT-6.2 Astra and takes much longer to complete. The listing says pro mode is intended for hard, high-stakes problems where the extra accuracy justifies the cost.

For everyday coding, agentic and chat workloads, the listing directs users to GPT-6.2 Astra instead. Details on the reasoning mode setting are in OpenAI's reasoning guide in its API documentation.

What this means

GPT-6.2 Astra Pro is a different way of running GPT-6.2 Astra, not a separate model. The difference is the amount of reasoning the model performs per request. The listed $8/$40 per 1M token rates are only part of the cost picture. Because pro mode uses far more reasoning tokens, the listing says a typical request costs several times more than on GPT-6.2 Astra, and responses take much longer.

Teams should reserve it for hard, high-stakes tasks, as the listing advises, and keep GPT-6.2 Astra for routine coding, agentic and chat workloads. Teams that need EU in-region routing should note it is not currently available for this model.

Comments

Loading...