GPT-6 Astra
OpenAI · released Sep 4, 2026
- Type
- Closed
- Input
- $10.00
- Output
- $50.00
- Cached
- $1.00
List price · per 1M tokens · OpenAI at 922K context · machine-readable source ↗
Our take
Written Sep 7, 2026GPT-6 Astra is OpenAI's flagship hosted model that can handle up to 1.05 million tokens in a single request, accepting text, images and files. It excels at mathematics and reasoning on refreshed competition problems, though its agentic coding and instruction following lag well behind its own peak scores.
Pick this for long-document analysis at one-million-plus tokens, mathematics workloads where it scores over 96%, or reasoning tasks at 92%. It is also the choice for web development given its Arena Code score. Skip it if you need strong agentic coding, where it drops 23 points below its standard coding score, or if you need consistent fast throughput — speed varies from 10 to 49 tokens per second depending on the host.
The case for it
- Exceptional mathematics on refreshed competition problems, at 96.81% LiveBench.
- Strong reasoning on monthly-refreshed tasks, at 92.65% LiveBench.
- One-million-token request limit, among the largest available.
- Entry tier from the same vendor costs a fraction of the premium tier.
The case against it
- Agentic coding lags 23 points behind its own standard coding score.
- Instruction following is its weakest sub-score, over 21 points below mathematics.
- Throughput varies sharply across hosts, from 10–14 to 35–49 tokens per second.
How good is it?
A closed text model for everyday questions, drafting, coding and multi-step agent work.
- getting answers to everyday questionsArena Text (overall) · 18th of 168
- drafts, rewrites and editingArena Creative Writing · 23rd of 168
- writing and completing codeArena Coding · 5th of 168
- multi-step work it carries out for youArena Agent · 2nd of 55
- calling tools to carry out requestsArena Agent · Tool use · 9th of 55
- getting back on track after a step failsArena Agent · Recovery · 11th of 55
EverydayGeneral questions and everyday reasoning
Arena Text (overall)18th of 168 · 1478
CodingWriting and fixing code on its own
Arena Coding5th of 168 · 1542
AgenticPlanning, calling tools, staying on task
Arena Agent2nd of 55 · 0.109
WritingDrafting and rewriting prose
Arena Creative Writing23rd of 168 · 1453
Placings on Arena's agent boards, from live sessions people ran themselves. A model can lead on one of these and sit mid-field on the others.
Boards this model appears on that none of the ratings above are built on.
Every published score for this model20 scoresEvery figure we hold, from 20 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Where to rent it
Prices checked 4 hours ago — each listing carries its own date.
Microsoft Azure AI, through OpenRouter
Why this differs from the header. The strip above quotes OpenAI's own list price; this is the cheapest live offer, whoever is serving it — a reseller undercutting a lab is ordinary commerce, not an error.
- per 1M tokens
- $10.00 in / $50.00 out
- Context served
- 1.1M
- Throughput
- ~24 tok/s
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| OpenAIflex tierThrough OpenRouter | $5.00 / $25.00checked 4 hours ago | 1.1M128K max reply | 25 tok/s | No | Yesunknown period | Unknown |
| Microsoft Azure AIThrough OpenRouter | $10.00 / $50.00checked 4 hours ago | 1.1M128K max reply | 24 tok/s | No | No | Confirmed |
| OpenRouterOpenRouter's own listing | $10.00 / $50.00checked 4 hours ago | 1.1M | not measured | Unknown | Unknown | Unknown |
| OpenAIDirect | $10.00 / $50.00checked 4 hours ago | 922K128K max reply | 35 tok/s | No | Yesunknown period | Unknown |
| Microsoft Azure AIusThrough OpenRouter | $11.00 / $55.00checked 4 hours ago | 1.1M128K max reply | 26 tok/s | No | No | Confirmed |
| Amazon Bedrockus-west-2Through OpenRouter | $11.00 / $55.00checked 4 hours ago | 1.1M128K max reply | 43 tok/s | No | No | Unknown |
| OpenAIfast tierThrough OpenRouter | $20.00 / $100.00checked 4 hours ago | 1.1M128K max reply | 42 tok/s | No | Yesunknown period | Unknown |
OpenRouter
Amazon BedrockAcross the 7 listings we hold: 6 say they do not train on prompts, 0 say they do and 1 does not say. 2 appear in the zero-retention registry we check; the rest are unknown to us.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| OpenAIflexThrough OpenRouter | ✓ | ✓ | ✓ |
| Microsoft Azure AIThrough OpenRouter | ✓ | ✓ | ✓ |
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✓ |
| OpenAIDirect | ✓ | ✓ | ✓ |
| Microsoft Azure AIusThrough OpenRouter | ✓ | ✓ | ✓ |
| Amazon Bedrockus-west-2Through OpenRouter | ✓ | ✓ | ✓ |
| OpenAIfastThrough OpenRouter | ✓ | ✓ | ✓ |
Tool calling: 7 of 7 listings say yes. JSON output: 7 of 7 listings say yes. Strict schema: 7 of 7 listings say yes.
When we formed this view
Recent changes
Each date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 1 of 7 listings does not say whether it trains on prompts.
- We hold no batch or off-peak rate for any of its listings.
Licence and identifiers
What the licence allowsWe hold no licence record for this model. Inside are the identifiers you need to pull it — its Hugging Face repo where we have one, our slug and a machine-readable card.
Licence
We hold no licence record for this model, and no record of published weights either — so we can neither summarise its terms nor point you at the weights.
Identifiers
- Takes in, gives back
- Text, images and documents in, text out
- Catalogue slug
- openai-gpt-6-astra