Back

OpenAI

GPT-6 Astra

ReasoningClosed weights
Intelligence
49.6
Rank 21 of 744 measured models
Coding
76.7
Rank 13 of 298 measured models
Output speed
43.8 t/s
Rank 346 of 403 measured models
Price/1M
$20
Rank 638 of 659 measured models
Context window
1M
Rank 84 of 1199 measured models
Cost per task
$1.54
Rank 151 of 193 measured models

GPT-6 Astra (Medium) scores 49.6 on the Artificial Analysis Intelligence Index, ranking 21st of 744 measured models. The closest models on this index are Claude Opus 5 (50.8), Claude Fable 5 (batch) (49.6), and Muse Spark 1.3 (48.1). At $20 per million tokens (3:1 input/output blended price), it is more expensive than Claude Opus 5 ($10) and Muse Spark 1.3 ($2). With an output speed of 43.8 t/s, it is slower than Claude Opus 5 (54.7 t/s) and Muse Spark 1.3 (128 t/s).

Artificial Analysis Indices
Medium
Intelligence49.6
Rank 21 of 744 measured models
Coding76.7
Rank 13 of 298 measured models
Agentic46.0
Rank 30 of 189 measured models
Reasoning levels
Each point is a reasoning level of this model: more reasoning often raises the score but costs more. Select a point to open that level.
Intelligence Indexmost efficient ↗Cost per task
Capability indexes
Performance in specific industries, measured by Artificial Analysis
Finance & Accounting52.2
Rank 21 of 218 measured models
Strategy & Ops54.2
Rank 25 of 218 measured models
Legal56.6
Rank 17 of 218 measured models
Engineering51.7
Rank 22 of 217 measured models
Economics57.8
Rank 20 of 225 measured models
Standard Benchmarks
GPQA Diamond93.9%
Rank 10 of 619 measured models
HLE52.7%
Rank 20 of 664 measured models
SciCode54.2%
Rank 63 of 231 measured models
AA-LCR79.7%
Rank 87 of 575 measured models
Terminal-Bench 2.189.5%
Rank 5 of 244 measured models
Terminal-Bench v449.5%
Rank 19 of 218 measured models
τ-Bench Banking35.5%
Rank 48 of 211 measured models
Models at a similar level
The closest models on the Intelligence Index, one configuration per family.

Preferences and analytics

Rybbit: visits, performance, clicks, copies, forms and session replay (masked inputs), with your permission. Refusing does not limit the site.

Privacy

Allow navigation analytics?

With your permission, Rybbit measures visits, performance and interactions (clicks, copies, forms), including session replay, to improve BenchSift. Input values are masked in replays. Refusing does not limit the site.

Your choice is stored for 180 days. Change it at any time using “Privacy preferences” in the footer. Withdrawal reloads the page.

No choice stored: Rybbit is disabled.

Privacy