By the toolist.ai editors. Researched from public sources, not hands-on tested by toolist.ai. Updated October 6, 2026; prices checked October 6, 2026.
GPT-6 Astra: price, context window and benchmarks
GPT-6 Astra is OpenAI's most capable GPT-6 model, at $10 per million input tokens and $50 per million output tokens, five times GPT-6.1 Sol. OpenAI aims it at the hardest reasoning, coding and research work; on Artificial Analysis it scores 52.7 on the Intelligence Index, a point above GPT-6.1 Sol.
Best for
- The hardest single tasks, where a small quality edge is worth five times the token price
- Long, hard terminal and coding tasks (59.1% on Terminal-Bench 4.0, the highest of the GPT-6 models)
- Science questions where it has a published GPQA Diamond score of 96.1%
Key facts
| Fact | GPT-6 Astra | Source |
|---|---|---|
| Provider | OpenAI | OpenAI API docs: Pricing |
| API model name | gpt-6-astra | OpenAI API docs: GPT-6 Astra |
| Released | September 3, 2026 | OpenAI API docs: Changelog |
| Input price | $10 per million tokens | OpenAI API docs: Pricing |
| Cached input price | $1 per million tokens | OpenAI API docs: Pricing |
| Output price | $50 per million tokens | OpenAI API docs: Pricing |
| Context window | 1,050,000 tokens | OpenAI API docs: GPT-6 Astra |
| Maximum output | 128,000 tokens | OpenAI API docs: GPT-6 Astra |
| Knowledge cutoff | April 30, 2026 | OpenAI API docs: GPT-6 Astra |
| Input and output | Text and images in; text out | OpenAI API docs: GPT-6 Astra |
| Reasoning | Always reasons. Effort levels low to max; "none" is not supported | OpenAI API docs: Reasoning models |
| API and cloud platforms | OpenAI API (Responses, Chat Completions and Batch) | OpenAI API docs: GPT-6 Astra |
| In the apps | ChatGPT Plus, Pro, Business, Enterprise and Edu, in ChatGPT Work and Codex; not on Free or Go | ChatGPT docs: Pricing and plan access |
Prompts over 272,000 input tokens are billed at twice the input rate and 1.5 times the output rate for the whole request. Batch and Flex cost half; Fast mode costs double and Ultrafast six times.
What GPT-6 Astra costs a month
Three typical workloads, priced at the standard API rates above. We calculated these; the provider does not publish them.
| Workload | Tokens a month | Cost a month |
|---|---|---|
| Support chatbot 100,000 replies, each 3,000 input and 400 output tokens, no caching | 300M in, 40M out | $5,000 |
| Document Q&A 5,000 questions over the same 60,000-token document set (90% cache hits), 800 output tokens each | 300M in, 4M out | $770 |
| Coding agent 200 tasks, each 1.5 million input tokens (90% cache hits) and 50,000 output tokens | 300M in, 10M out | $1,070 |
Reasoning models bill their thinking as output tokens, so heavy reasoning can raise the output count several times. Cache-write charges are not included.
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| Artificial Analysis Intelligence Index max effort | 52.7 | Artificial Analysis: GPT-6 Astra (max) |
| Intelligence Index at everyday effort medium effort; OpenAI publishes no default | 49.6 | Artificial Analysis: GPT-6 Astra (medium) |
| Coding Agent Index Codex harness, max effort | 61.6 | Artificial Analysis: Coding agents |
| Terminal-Bench 4.0 max effort | 59.1% | Artificial Analysis: GPT-6.1 Sol vs GPT-6 Astra |
| Humanity's Last Exam max effort | 54.7% | Artificial Analysis: GPT-6.1 Sol vs GPT-6 Astra |
| GDPval-AA v2.1 (Elo) max effort | 1542 | Artificial Analysis: GPT-6.1 Sol vs GPT-6 Astra |
| SciCode max effort | 56.5% | Artificial Analysis: GPT-6.1 Sol vs GPT-6 Astra |
| Terminal-Bench-Science 0.1 max effort | 63.3% | Artificial Analysis: GPT-6 Astra (max) |
| GPQA Diamond max effort | 96.1% | Artificial Analysis: GPT-6 Astra (max) |
Speed: about 50.9 output tokens a second, 5.5 seconds to the first token (medium effort, OpenAI API; Artificial Analysis: GPT-6 Astra (medium)).
Benchmark and speed data marked as such come from Artificial Analysis.
What changed
The first GPT-6 model and a new top tier above Sol. OpenAI's guide describes Astra as its highest-intelligence model, Sol as balanced between speed, cost and intelligence, and Luna as the fastest and cheapest. (OpenAI API docs: Using GPT-6)
GPT-6 Astra compared
Other models
- GPT-6.1 Sol: $2 in, $10 out per million tokens
- Claude Opus 5.5: $4 in, $20 out per million tokens
- GPT-6 Sol: $2 in, $10 out per million tokens
- Claude Fable 5.1: $10 in, $50 out per million tokens
- Claude Opus 5: $5 in, $25 out per million tokens
All prices side by side: AI model API price table. How we research: our method.
Sources
- OpenAI API docs: GPT-6 Astra (checked 2026-10-06)
- OpenAI API docs: Pricing (checked 2026-10-06)
- OpenAI API docs: Changelog (checked 2026-10-06)
- OpenAI API docs: Using GPT-6 (checked 2026-10-06)
- OpenAI API docs: Reasoning models (checked 2026-10-06)
- ChatGPT docs: Models (checked 2026-10-06)
- ChatGPT docs: Pricing and plan access (checked 2026-10-06)
- Artificial Analysis: GPT-6 Astra (max) (checked 2026-10-06)
- Artificial Analysis: GPT-6 Astra (medium) (checked 2026-10-06)
- Artificial Analysis: GPT-6.1 Sol vs GPT-6 Astra (checked 2026-10-06)
- Artificial Analysis: Coding agents (checked 2026-10-06)
- OpenRouter: GPT-6 Astra (checked 2026-10-06)