Models/Gemma 3n E4B Instruct

MDL

Gemma 3n E4B Instruct

Google

Verified against Artificial Analysis · Jul 13, 2026

Capability

percentile index

Price /M

$0.02 / $0.04

in / out

Context

—

Avg score

21.1

11 benchmarks

AA Intelligence

1.2

index

Speed

56 tok/s

TTFT 0.4s

Released

Jun 26, 2025

About

Gemma 3n E4B Instruct is a open-weights model in the Gemma family from Google. Benchmarked on 11 evals, averaging 21.1.

Benchmark scores · 11

avg 21.1 — higher is better

MATHartificial-analysis · 2026-07

77.1

MMLU-Proartificial-analysis · 2026-07

48.8

GPQA Diamondartificial-analysis · 2026-07

29.6

IFBenchartificial-analysis · 2026-07

27.9

LiveCodeBenchartificial-analysis · 2026-07

14.6

AIMEartificial-analysis · 2026-07

14.3

SciCodeartificial-analysis · 2026-07

8.1

τ²-benchartificial-analysis · 2026-07

Humanity's Last Examartificial-analysis · 2026-07

4.4

Terminal-Bench Hardartificial-analysis · 2026-07

2.3

AA Long-Context Reasoningartificial-analysis · 2026-07

Details

PublisherGoogle

FamilyGemma

Parameters—

License—

WeightsOpen

ReleasedJun 26, 2025

Performance & price

Output speed56 tok/s

Latency (TTFT)0.4s

Price in / out$0.02 / $0.04 per 1M

Blended price$0.025 / 1M

Coding index3.2

Math index14.3

Median across API providers, via Artificial Analysis. Blended = 3:1 input:output. For reasoning models, latency includes thinking time. Methodology →

History

full ledger →

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers

Artificial Analysis$0.04/M