Gemma 4 E2B (Reasoning)

Gemma 4 E2B (Reasoning)

Google · Released Apr 2, 2026 Emerging
Intelligence #489 / 583
18.5 Our Score
AA Index #233 / 380
9.3 Artificial Analysis
Input
— Not priced
Output
— Not priced
Context
— Not reported

Analysis Summary

Gemma 4 E2B (Reasoning) is the reasoning-enabled variant of Google's smallest Gemma 4 model, with an intelligence index of 9.3 and coding and agentic indices of 9 and 11.9. The reasoning mode provides a marginal uplift over the non-reasoning sibling on GPQA (0.433 vs 0.405) but overall capability remains very limited.

For businesses, this model is not suitable for production workloads. Its instruction-following, long-context, and coding scores are all well below the thresholds required for reliable business use. It may serve as a research baseline or edge-deployment prototype.

No pricing data is available. Teams evaluating this model should consider whether the reasoning variant adds meaningful value over the non-reasoning sibling for their specific constrained use case, as the capability gap versus larger models is substantial.

Assessed June 19, 2026

Editorial notes

Gemma 4 E2B (Reasoning) from Google adds a reasoning mode to the smallest E2B base but remains limited across coding, agentic, and instruction-following dimensions for business use.

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?

Performance Profile

Intelligence1.7Technical1.2Value0Content3
Intelligence 1.7/10
Technical 1.2/10
Content 3/10
Value 0/10

How Gemma 4 E2B (Reasoning) compares

Gemma 4 E2B (Reasoning) ranks #233 of 380 AI models we track for overall intelligence, #248 of 292 for agentic tasks. Gemma 4 E2B (Reasoning) is currently free to use via OpenRouter.

Performance Indices

Source: Artificial Analysis

9.3 Intelligence Index
11.9 Agentic Index

Benchmark Scores

Intelligence

GPQA Diamond 43.3% Graduate-level scientific reasoning
HLE 4.8% Humanity's Last Exam
SciCode 20.9% Scientific computing

Technical

TerminalBench Hard 3% Agentic terminal tasks
τ²-Bench 20.8% Conversational agent benchmark

Content

IFBench 38% Instruction following
LCR 15% Long-context reasoning

Benchmark data from Artificial Analysis and Hugging Face

How does Gemma 4 E2B (Reasoning) stack up?

Compare side-by-side with other emerging models.

Compare Models

Model Information

ProviderGoogle
Release Date April 2, 2026
Status Active

Frequently asked questions about Gemma 4 E2B (Reasoning)

How much does Gemma 4 E2B (Reasoning) cost?

Gemma 4 E2B (Reasoning) is currently available for free via OpenRouter.

Who created Gemma 4 E2B (Reasoning)?

Gemma 4 E2B (Reasoning) is developed by Google and was released on April 2, 2026.