Back

OpenAI

GPT-5.6 Terra

ReasoningClosed weights
Intelligence
38.0
Rank 80 of 744 measured models
Coding
70.6
Rank 54 of 298 measured models
Output speed
98.4 t/s
Rank 172 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
$0.632
Rank 113 of 193 measured models

GPT-5.6 Terra (Xhigh) scores 38.0 on the Artificial Analysis Intelligence Index, ranking 80th of 744 measured models. The closest models on this index are Claude Sonnet 5 (38.2), GPT-6 Luna (38.1), and MiMo-V2.6-Flash (37.9). At $4.50 per million tokens (3:1 input/output blended price), it is more expensive than Claude Sonnet 5 ($4), GPT-6 Luna ($0.20), and MiMo-V2.6-Flash ($0.175). With an output speed of 98.4 t/s, it is faster than Claude Sonnet 5 (74.7 t/s) and MiMo-V2.6-Flash (53.8 t/s), slower than GPT-6 Luna (117 t/s).

Artificial Analysis Indices
Extra high
Intelligence38.0
Rank 80 of 744 measured models
Coding70.6
Rank 54 of 298 measured models
Agentic41.4
Rank 40 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 & Accounting39.8
Rank 78 of 218 measured models
Strategy & Ops43.5
Rank 71 of 218 measured models
Legal39.8
Rank 84 of 218 measured models
Healthcare & Medical37.6
Rank 38 of 82 measured models
Engineering37.8
Rank 76 of 217 measured models
Economics44.1
Rank 86 of 225 measured models
Standard Benchmarks
GPQA Diamond90.8%
Rank 57 of 619 measured models
HLE41.9%
Rank 73 of 664 measured models
SciCode52.3%
Rank 83 of 231 measured models
IFBench66.3%
Rank 98 of 450 measured models
AA-LCR79.0%
Rank 104 of 575 measured models
TerminalBench Hard62.9%
Rank 2 of 432 measured models
Terminal-Bench 2.180.1%
Rank 46 of 244 measured models
Terminal-Bench v410.1%
Rank 87 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