Back

OpenAI

GPT-5.6 Terra

ReasoningClosed weightsModels.dev
Intelligence
42.1
Rank 52 of 744 measured models
Coding
76.7
Rank 13 of 298 measured models
Output speed
107 t/s
Rank 159 of 403 measured models
Price/1M
$4.50
Rank 563 of 659 measured models
Context window
1M
Rank 84 of 1199 measured models
Cost per task
$1.40
Rank 148 of 193 measured models

GPT-5.6 Terra (Max) scores 42.1 on the Artificial Analysis Intelligence Index, ranking 52nd of 744 measured models. The closest models on this index are Claude Opus 4.8 (batch) (41.8), GLM-5.3-Flash (41.8), and Ling 3.1 Flash (41.1). At $4.50 per million tokens (3:1 input/output blended price), it is more expensive than GLM-5.3-Flash ($0.237) and Ling 3.1 Flash ($0.45). With an output speed of 107 t/s, it is faster than GLM-5.3-Flash (61.1 t/s), slower than Ling 3.1 Flash (211 t/s).

Artificial Analysis Indices
Maximum
Intelligence42.1
Rank 52 of 744 measured models
Coding76.7
Rank 13 of 298 measured models
Agentic43.2
Rank 35 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 & Accounting41.4
Rank 66 of 218 measured models
Strategy & Ops47.5
Rank 53 of 218 measured models
Legal41.5
Rank 75 of 218 measured models
Healthcare & Medical40.3
Rank 32 of 82 measured models
Engineering43.1
Rank 51 of 217 measured models
Economics44.9
Rank 82 of 225 measured models
Standard Benchmarks
GPQA Diamond92.5%
Rank 36 of 619 measured models
HLE42.9%
Rank 59 of 664 measured models
SciCode55.0%
Rank 52 of 231 measured models
IFBench71.2%
Rank 59 of 450 measured models
AA-LCR83.0%
Rank 25 of 575 measured models
TerminalBench Hard57.6%
Rank 11 of 432 measured models
Terminal-Bench 2.188.0%
Rank 12 of 244 measured models
Terminal-Bench v435.4%
Rank 33 of 218 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