Drafted with AI tools and edited by a human. The figures and conclusions were checked by a GPUniq editor.
GPT-6.2 Astra Pricing and Specs
The GPT-6.2 Astra API charges $8.00 per million standard input tokens. Cached input tokens drop to $0.40 per million, and output tokens cost $40.00 per million. Released on November 17, 2026, OpenAI's reasoning model supports a 1,050,000 token context window and a 128,000 token maximum output limit.
| Token Type | Price, USD per million tokens |
|---|---|
| Standard input | 8.00 |
| Cached input | 0.40 |
| Output | 40.00 |
Benchmark Performance
Performance beats the predecessor GPT-6 Astra on scientific evaluations. OpenAI reports a 5.2 point improvement on Agents' Last Exam and a 9 percentage point accuracy gain on Terminal-Bench-Science 0.1.
| Benchmark | Score Comparison against GPT-6 Astra |
|---|---|
| Agents' Last Exam | 5.2 points higher |
| Terminal-Bench-Science 0.1 | 9 percentage points higher |
Modalities and Endpoints
Inputs are restricted to text and images. Audio and video inputs are not supported, and fine-tuning is unavailable.
Access the model via OpenAI API endpoints including Chat Completions, Responses, Real-time, and Batch endpoints. It is also available through Azure AI Foundry and features ultrafast mode at launch. Users can access it in ChatGPT Work and Codex across Plus, Pro, Enterprise, and Edu tiers.
Related on the GPUniq blog
Methodology
Facts are compiled from the vendor's own announcement and documentation, linked below, at the time of writing; where the vendor hasn't published a number it is marked as such. GPUniq prices, when the model is already available here, come from our live catalog.
Sources
- 1.GPT-6.2 Astra Model | OpenAI API — OpenAI (accessed )
- 2.Introducing GPT-6.2 Astra and Sol — OpenAI (accessed )
- 3.GPT-6.2 System Card - OpenAI Deployment Safety Hub — OpenAI (accessed )
Want cheap GPUs for your next project?
Browse live GPU prices and rent the right card in seconds — H100, A100, RTX 4090, and 50+ more models.
