GPT-6 Astra
Input above 272K tokens is billed 2x on input and 1.5x on output. Offensive cybersecurity capability is withheld and released only through controlled channels; fine-tuning is not supported.
01
Snapshot
Overall
96
rank #1 of 26
Coding
97
rank #3 of 26
Multimodal
90
rank #7 of 26
Input / Output
$10 / $50
USD per 1M tokens
Context
1.05M
usable recall 96
Speed
~35 tok/s
TTFT 1.1s
Aggregated from public sources and independently weighted; methodology on the Terms page. Scores within 3 points are treated as statistically tied.
02
Specs & pricing
Every tracked field. List API prices in USD per 1M tokens; verify current vendor pricing before purchase.
| Vendor | OpenAI (US) |
| Released / current | 2026.09 |
| Licensing | Closed / managed API |
| Context window | 1.05M |
| Max output | 128K |
| Effective-context score | 96 |
| Input / output price | $10 / $50 per 1M tokens |
| Cache discount | Prompt cache 50% off |
| Free tier | No free API tier; Fast mode delivers up to ~2x throughput |
| Speed / TTFT | ~35 tok/s / 1.1s |
| Function calling | 94 |
| Refusal rate | ~13% |
| English / Chinese | 99 / 72 |
| Modalities | text, image |
| Fine-tuning | No |
| Private deployment | No |
| SOC2 / no-train | yes / yes |
03
Strengths & watch-outs
Strengths
- Best-in-class computer use
- Frontier research & reasoning
- Autonomous long-horizon tasks
- Strongest cybersecurity (gated)
Watch-outs
- Highest price ($10/$50)
- No native audio/video
- Coding only neck-and-neck
- Critical capability is gated
04
Where it ranks among 26 models
Overall 96/100 (#1), coding 97 (#3), multimodal 90 (#7). See the full boards and side-by-side compare on the leaderboard.
05
Pricing in practice
Illustrative monthly bill for 100M input + 30M output tokens: $2,500 at list; about $2,050 with 90% of inputs cache-hit. Your mix and cache-hit ratio change this.
Illustrative model using public list rates before any enterprise agreement; cache discount: Prompt cache 50% off.
06
Frequently asked questions
How much does GPT-6 Astra cost?
List API pricing is $10 input and $50 output per 1M tokens. Prompt cache 50% off. No free API tier; Fast mode delivers up to ~2x throughput
What is GPT-6 Astra's context window, and is it usable end to end?
It advertises a 1.05M window with 128K max output; its effective-context score is 96/100, which is the better predictor of whether long-document details are actually retained.
Is GPT-6 Astra good for coding?
Its coding aggregate is 97/100, rank #3 of 26. That puts it in the top tier for agentic, multi-file engineering.
Can I self-host GPT-6 Astra?
No. It is a closed managed API; there is no weights download.
How strong is it in Chinese and on multimodal inputs?
Chinese score is 72 versus English 99; multimodal aggregate is 90/100 (rank #7), accepting text, image.