Back

OpenAI

GPT-5.4 mini

ReasoningClosed weightsModels.dev
Intelligence
24.1
Rank 189 of 744 measured models
Coding
56.1
Rank 101 of 298 measured models
Output speed
233 t/s
Rank 24 of 403 measured models
Price/1M
$1.69
Rank 448 of 659 measured models
Context window
400K
Rank 357 of 1199 measured models
Cost per task
$0.447
Rank 98 of 193 measured models

GPT-5.4 mini (Xhigh) scores 24.1 on the Artificial Analysis Intelligence Index, ranking 189th of 744 measured models. The closest models on this index are DeepSeek V4 Flash 0420 (24.2), Solar Mini 4 (24.1), and MiMo-V2-Omni (23.9). At $1.69 per million tokens (3:1 input/output blended price), it is more expensive than DeepSeek V4 Flash 0420 ($0.142) and Solar Mini 4 ($0.175). With an output speed of 233 t/s, it is faster than DeepSeek V4 Flash 0420 (39.6 t/s) and Solar Mini 4 (83.7 t/s).

Artificial Analysis Indices
Extra high
Intelligence24.1
Rank 189 of 744 measured models
Coding56.1
Rank 101 of 298 measured models
Agentic17.9
Rank 101 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 ↗Price / 1M tokens
Capability indexes
Performance in specific industries, measured by Artificial Analysis
Finance & Accounting24.2
Rank 139 of 218 measured models
Strategy & Ops25.3
Rank 128 of 218 measured models
Legal26.2
Rank 140 of 218 measured models
Healthcare & Medical22.1
Rank 68 of 82 measured models
Engineering24.4
Rank 142 of 217 measured models
Economics29.7
Rank 156 of 225 measured models
Standard Benchmarks
GPQA Diamond87.5%
Rank 101 of 619 measured models
HLE28.1%
Rank 168 of 664 measured models
SciCode52.1%
Rank 84 of 231 measured models
IFBench73.3%
Rank 43 of 450 measured models
AA-LCR77.0%
Rank 137 of 575 measured models
TerminalBench Hard52.3%
Rank 19 of 432 measured models
Terminal-Bench 2.159.2%
Rank 106 of 244 measured models
Terminal-Bench v42.0%
Rank 114 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