Back

OpenAI

GPT-5.6 Terra

ReasoningClosed weights
Intelligence
27.5
Rank 148 of 744 measured models
Coding
58.1
Rank 96 of 298 measured models
Output speed
93.2 t/s
Rank 188 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
$0.145
Rank 64 of 193 measured models

GPT-5.6 Terra (Low) scores 27.5 on the Artificial Analysis Intelligence Index, ranking 148th of 744 measured models. The closest models on this index are Gemini 3 Pro Preview (28.0), GLM-5 (27.9), and Grok Build 0.1 0616 (27.2). At $4.50 per million tokens (3:1 input/output blended price), it is priced close to Gemini 3 Pro Preview ($4.50), more expensive than GLM-5 ($1.55) and Grok Build 0.1 0616 ($1.25). With an output speed of 93.2 t/s, it is faster than GLM-5 (71 t/s) and Grok Build 0.1 0616 (67.4 t/s).

Artificial Analysis Indices
Low
Intelligence27.5
Rank 148 of 744 measured models
Coding58.1
Rank 96 of 298 measured models
Agentic25.9
Rank 82 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 & Accounting29.5
Rank 115 of 218 measured models
Strategy & Ops31.3
Rank 112 of 218 measured models
Legal30.6
Rank 117 of 218 measured models
Healthcare & Medical27.5
Rank 57 of 82 measured models
Engineering27.5
Rank 126 of 217 measured models
Economics34.4
Rank 135 of 225 measured models
Standard Benchmarks
GPQA Diamond84.3%
Rank 151 of 619 measured models
HLE29.2%
Rank 158 of 664 measured models
SciCode49.9%
Rank 111 of 231 measured models
IFBench59.7%
Rank 130 of 450 measured models
AA-LCR71.3%
Rank 207 of 575 measured models
TerminalBench Hard43.9%
Rank 37 of 432 measured models
Terminal-Bench 2.162.5%
Rank 93 of 244 measured models
Terminal-Bench v41.5%
Rank 118 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