Back

Google

Gemma 4 E4B

ReasoningOpen weightsHugging Face
Intelligence
8.9
Rank 470 of 744 measured models
Coding
9.4
Rank 276 of 298 measured models
Output speed
41.7 t/s
Rank 357 of 403 measured models
Price/1M
$0.040
Rank 42 of 659 measured models
Context window
128K
Rank 773 of 1199 measured models
Cost per task
$0.0082
Rank 11 of 193 measured models

Gemma 4 E4B (Reasoning) scores 8.9 on the Artificial Analysis Intelligence Index, ranking 470th of 744 measured models. The closest models on this index are Devstral Medium (9.0), Claude 3.5 Haiku (8.9), and Gemini 2.0 Flash (Feb '25) (8.9).

Artificial Analysis Indices
Reasoning
Intelligence8.9
Rank 470 of 744 measured models
Coding9.4
Rank 276 of 298 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 & Accounting7.6
Rank 202 of 218 measured models
Strategy & Ops3.2
Rank 215 of 218 measured models
Legal11.4
Rank 188 of 218 measured models
Engineering14.8
Rank 186 of 217 measured models
Economics17.9
Rank 190 of 225 measured models
Standard Benchmarks
GPQA Diamond57.6%
Rank 423 of 619 measured models
HLE3.8%
Rank 593 of 664 measured models
SciCode24.4%
Rank 219 of 231 measured models
IFBench44.2%
Rank 228 of 450 measured models
AA-LCR32.0%
Rank 400 of 575 measured models
TerminalBench Hard8.3%
Rank 249 of 432 measured models
Terminal-Bench 2.11.9%
Rank 229 of 244 measured models
Terminal-Bench v40.0%
Rank 151 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