GeekRadar / The comparison desk

COMPARE.

AI models

google

Gemma 3 4B IT

AI model Released Mar 12, 2025

API mode: Chat^*
Context window: 131,072 tokens
Maximum output: 131,072 tokens
Open weights: Yes

Capability & preference

Human preference · Arena

Overall text preferences from head-to-head votes. Not factual accuracy or a task-success rate.

Gemma 3 4B IT 1290.8 Gemma 3 4B IT · 95% interval 1281.6–1300.0 · 4,171 votes · 2026-10-28

Model performance

Radar task score: four tests, equal weight. An experimental summary of these tasks, not a universal intelligence rating.

Gemma 3 4B IT

Tested configuration Radar task score

Unrated

1/4 core tests · not enough evidence

2 benchmark results for this configuration

Scientific reasoning

GPQA Diamond · 25%

Expert-written science questions; strict multiple-choice scoring. Chance accuracy is 25%.

Gemma 3 4B IT 23.2^*% Approx. 95% interval: 19.6–26.9% · 2026-08-28

Novel problem solving

Mystery Game Puzzles · 25%

Best-next-move puzzles in an undisclosed game variant. A narrow reasoning test, not general intelligence.

Gemma 3 4B IT Not available

Advanced mathematics

FrontierMath Tiers 1-3 · 25%

Private v2 mathematics problems with Python tools. Earlier versions and Tier 4 are excluded.

Gemma 3 4B IT Not available

Factual answers

SimpleQA Verified · 25%

Correct factual answers under a guessing prompt, graded by a model. Not a hallucination or abstention score.

Gemma 3 4B IT Not available

Competition mathematics

OTIS Mock AIME 2024-2025 · Separate result

Competition problems with model-extracted answers checked against an exact key. Earlier grading procedures are excluded.

Gemma 3 4B IT 7.5^*% Approx. 95% interval: 2.1–12.9% · 2026-08-28

Bars show reported task scores. Whiskers are approximate 95% intervals where standard errors are available. Different reasoning budgets are not equal-cost tests. Missing results are never zero.

Recorded results

Only the same benchmark, version, workload, scenario and unit share a row. Missing results are not zero. System results include the full submitted hardware and software configuration.

Test version

Arena text preference / codinghuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,229.96 rating

Confidence interval: 1,206.59 – 1,253.33

605 votes

Oct 28, 2026

Category: coding
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / creative_writinghuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,270.14 rating

Confidence interval: 1,248.35 – 1,291.92

666 votes

Oct 28, 2026

Category: creative_writing
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / englishhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,300.39 rating

Confidence interval: 1,288.86 – 1,311.92

2572 votes

Oct 28, 2026

Category: english
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / exclude_tieshuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,220.92 rating

Confidence interval: 1,207.14 – 1,234.70

2830 votes

Oct 28, 2026

Category: exclude_ties
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / experthuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,222.84 rating

Confidence interval: 1,182.20 – 1,263.47

208 votes

Oct 28, 2026

Category: expert
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / germanhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,280.66 rating

Confidence interval: 1,238.53 – 1,322.78

162 votes

Oct 28, 2026

Category: german
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / hard_promptshuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,252.51 rating

Confidence interval: 1,235.47 – 1,269.55

1092 votes

Oct 28, 2026

Category: hard_prompts
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / hard_prompts_englishhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,253.22 rating

Confidence interval: 1,232.96 – 1,273.47

764 votes

Oct 28, 2026

Category: hard_prompts_english
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,310.89 rating

Confidence interval: 1,280.86 – 1,340.92

390 votes

Oct 28, 2026

Category: industry_business_and_management_and_financial_operations
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,232.72 rating

Confidence interval: 1,212.42 – 1,253.02

761 votes

Oct 28, 2026

Category: industry_entertainment_and_sports_and_media
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,327.24 rating

Confidence interval: 1,291.51 – 1,362.97

245 votes

Oct 28, 2026

Category: industry_legal_and_government
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,305.48 rating

Confidence interval: 1,284.87 – 1,326.09

780 votes

Oct 28, 2026

Category: industry_life_and_physical_and_social_science
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / industry_mathematicalhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,255.97 rating

Confidence interval: 1,226.10 – 1,285.84

365 votes

Oct 28, 2026

Category: industry_mathematical
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,341.72 rating

Confidence interval: 1,303.51 – 1,379.93

197 votes

Oct 28, 2026

Category: industry_medicine_and_healthcare
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,268.58 rating

Confidence interval: 1,250.21 – 1,286.95

986 votes

Oct 28, 2026

Category: industry_software_and_it_services
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,288.02 rating

Confidence interval: 1,271.08 – 1,304.96

1080 votes

Oct 28, 2026

Category: industry_writing_and_literature_and_language
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / instruction_followinghuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,238.85 rating

Confidence interval: 1,222.70 – 1,255.00

1233 votes

Oct 28, 2026

Category: instruction_following
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / longer_queryhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,273.23 rating

Confidence interval: 1,246.28 – 1,300.19

444 votes

Oct 28, 2026

Category: longer_query
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / mathhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,239.15 rating

Confidence interval: 1,211.21 – 1,267.08

423 votes

Oct 28, 2026

Category: math
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / multi_turnhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,255.17 rating

Confidence interval: 1,231.77 – 1,278.57

565 votes

Oct 28, 2026

Category: multi_turn
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / non_englishhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,273.10 rating

Confidence interval: 1,259.16 – 1,287.04

1599 votes

Oct 28, 2026

Category: non_english
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / overallhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,290.78 rating

Confidence interval: 1,281.55 – 1,300.00

4171 votes

Oct 28, 2026

Category: overall
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / russianhuman preference · 2026-10-28 · rating

Gemma 3 4B IT

1,293.80 rating

Confidence interval: 1,263.36 – 1,324.24

307 votes

Oct 28, 2026

Category: russian
Evaluated configuration: Gemma 3 4B IT
Evaluation type: human preference

Arena text preference / codinghuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,230.05 rating

Confidence interval: 1,206.68 – 1,253.43

605 votes

Oct 15, 2026

Category: coding
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / creative_writinghuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,270.31 rating

Confidence interval: 1,248.52 – 1,292.10

666 votes

Oct 15, 2026

Category: creative_writing
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / englishhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,300.36 rating

Confidence interval: 1,288.83 – 1,311.89

2572 votes

Oct 15, 2026

Category: english
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / exclude_tieshuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,220.96 rating

Confidence interval: 1,207.18 – 1,234.74

2830 votes

Oct 15, 2026

Category: exclude_ties
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / experthuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,222.84 rating

Confidence interval: 1,182.19 – 1,263.49

208 votes

Oct 15, 2026

Category: expert
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / germanhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,280.59 rating

Confidence interval: 1,238.47 – 1,322.70

162 votes

Oct 15, 2026

Category: german
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / hard_promptshuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,252.52 rating

Confidence interval: 1,235.47 – 1,269.56

1092 votes

Oct 15, 2026

Category: hard_prompts
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / hard_prompts_englishhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,253.12 rating

Confidence interval: 1,232.87 – 1,273.38

764 votes

Oct 15, 2026

Category: hard_prompts_english
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,310.95 rating

Confidence interval: 1,280.91 – 1,340.99

390 votes

Oct 15, 2026

Category: industry_business_and_management_and_financial_operations
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,232.78 rating

Confidence interval: 1,212.47 – 1,253.08

761 votes

Oct 15, 2026

Category: industry_entertainment_and_sports_and_media
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,326.99 rating

Confidence interval: 1,291.25 – 1,362.72

245 votes

Oct 15, 2026

Category: industry_legal_and_government
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,305.50 rating

Confidence interval: 1,284.89 – 1,326.11

780 votes

Oct 15, 2026

Category: industry_life_and_physical_and_social_science
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / industry_mathematicalhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,255.99 rating

Confidence interval: 1,226.12 – 1,285.87

365 votes

Oct 15, 2026

Category: industry_mathematical
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,341.81 rating

Confidence interval: 1,303.58 – 1,380.03

197 votes

Oct 15, 2026

Category: industry_medicine_and_healthcare
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,268.66 rating

Confidence interval: 1,250.28 – 1,287.03

986 votes

Oct 15, 2026

Category: industry_software_and_it_services
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,288.08 rating

Confidence interval: 1,271.13 – 1,305.02

1080 votes

Oct 15, 2026

Category: industry_writing_and_literature_and_language
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / instruction_followinghuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,238.84 rating

Confidence interval: 1,222.69 – 1,254.99

1233 votes

Oct 15, 2026

Category: instruction_following
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / longer_queryhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,273.30 rating

Confidence interval: 1,246.33 – 1,300.26

444 votes

Oct 15, 2026

Category: longer_query
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / mathhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,239.17 rating

Confidence interval: 1,211.23 – 1,267.11

423 votes

Oct 15, 2026

Category: math
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / multi_turnhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,255.20 rating

Confidence interval: 1,231.80 – 1,278.60

565 votes

Oct 15, 2026

Category: multi_turn
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / non_englishhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,273.19 rating

Confidence interval: 1,259.25 – 1,287.13

1599 votes

Oct 15, 2026

Category: non_english
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / overallhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,290.78 rating

Confidence interval: 1,281.56 – 1,300.01

4171 votes

Oct 15, 2026

Category: overall
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Arena text preference / russianhuman preference · 2026-10-15 · rating

Gemma 3 4B IT

1,293.91 rating

Confidence interval: 1,263.47 – 1,324.36

307 votes

Oct 15, 2026

Category: russian
Evaluated configuration: gemma-3-4b-it
Evaluation type: human preference

Run it locally

Inference · memory estimate

Gemma 3 4B IT

8.0 GiB checkpoint tensor payload · 4.30 B stored tensor elements · BF16

Stored tensors can include buffers and repeated shared weights. A verified unique parameter count is unavailable. Checkpoint size is not the memory required to run the model.

API cost

USD / month

Input tokens / monthOutput tokens / month

Standard text-token estimate. Excludes taxes, caching, batch discounts, reasoning and multimodal surcharges. Provider limits and regional availability may differ.

Gemma 3 4B IT

Provider offer

Gemma 3 4B IT / Pricing tiers 1

Provider / scope	Rate	Currency	Amount / unit
DeepInfragoogle/gemma-3-4b-itStandard	Input	USD	0.05 / 1M tokens
DeepInfragoogle/gemma-3-4b-itStandard	Output	USD	0.1 / 1M tokens

Capabilities

Context window: 131,072 tokens
Maximum input: 128,000 tokens^*
Maximum output: 131,072 tokens
Reasoning: No
Tool calling: Yes
Open weights: Yes
Knowledge cutoff: 2024-08
API mode: Chat^*
Image input: Yes^*
Audio input: No^*
System messages: Yes^*
Model license: Gemma
Checkpoint tensor payload: 8.01 GiB

Input modalities

text
image

Output modalities

text

System configuration

Detailed specifications

Specifications · 18 fields

deprecation date

2027-03-30^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

input cost per token

4.0e-8^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

input cost per token batches

2.0e-8^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

input cost per token flex

2.0e-8^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

input cost per token priority

7.0e-8^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

litellm provider

bedrock_converse^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

max input tokens

128,000^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

max output tokens

8,000^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

max tokens

8,000^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

mode

chat^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

output cost per token

8.0e-8^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

output cost per token batches

4.0e-8^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

output cost per token flex

4.0e-8^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

output cost per token priority

1.4e-7^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

supports audio input

No^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

supports function calling

Yes^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

supports system messages

Yes^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

supports vision

Yes^*

Scope and conditions

native_context

provider: bedrock_converse^*
mode: chat^*
full_native_model_id: google.gemma-3-4b-it^*
pricing_scope: Literal source field names and values retained; provider, region, cache/batch/threshold conditions and per-unit meaning must be checked. max_tokens is not automatically a context limit.^*

* Provisional data