Google · Text

Gemma models: Versions, specs, GPU requirements & hosting

Google Gemma separates full-size general models from edge-oriented variants and keeps model sizes as siblings within a generation.

Current releases Browse engineering services Compare models

Technical deployment reference illustration for Gemma — Reference deployment flow; exact architecture facts are listed separately.

Overview and history

Gemma is an official Google model family represented here through independently verified release, checkpoint, license, runtime and hardware evidence. Google Gemma separates full-size general models from edge-oriented variants and keeps model sizes as siblings within a generation. Current public releases are grouped by the General, Edge tracks so sibling sizes and specialist branches are not presented as false chronological generations.

FusionPointAI currently tracks 23 certified public Gemma releases. The catalogue preserves release dates and evaluates Current, Recent, Previous and Legacy status within each model track rather than by database insertion order.

Compact through large sizes

Edge track

Multimodal variants

Architecture and model range

The verified public range spans 5.1B to 31.3B total parameters and context windows up to 262,144 tokens. Each release states whether it is dense or mixture-of-experts and records active parameters separately when the official source publishes them.

Public releases23

Model tracks2

Current generationGemma 4 12B IT

Last verified2026-09-03T00:00:00.000000Z

Generations, tracks and variants

General

gemma-4-26b-a4b

Recent generation

Gemma-4-26B-A4B-NVFP4

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Quantization

gemma-4-31b-it

Recent generation

Gemma-4-31B-IT-NVFP4

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Quantization Unknown / Not Recommendable

Gemma 3

Recent generation 2 sibling variants

Gemma 3 12B IT

12.0B parameters · 128,000 context

Dense Dense Transformer

Gemma 3 27B IT

27.0B parameters · 128,000 context

Dense Dense Transformer

contextpilot-e4b

Recent generation

ContextPilot-E4B

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Fine-tune Unknown / Not Recommendable

gemma-4

Current generation 15 sibling variants

gemma-4-E4B-it-qat-q4_0-unquantized

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Fine-tune

gemma-4-E2B-it-qat-q4_0-unquantized

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Fine-tune

gemma-4-26B-A4B-it-qat-q4_0-unquantized

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Fine-tune

gemma-4-12B-it-qat-q4_0-unquantized

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Fine-tune

gemma-4-E2B-it-qat-mobile-transformers

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Quantization

gemma-4-31B-it-qat-q4_0-unquantized

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Fine-tune

gemma-4-E2B-it

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Base

gemma-4-E4B-it-qat-mobile-transformers

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Quantization

gemma-4-E4B-it

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Base

gemma-4-26B-A4B-it

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Base

gemma-4-12B-it

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Base

gemma-4-31B-it

The manufacturer has not published this information. parameters · N/A / Not directly comparable context

Base

Gemma 4 26B-A4B IT

25.8B parameters · 262,144 context

MoE Gemma4ForConditionalGeneration

Gemma 4 31B IT

31.3B parameters · 262,144 context

Dense Gemma4ForConditionalGeneration

Gemma 4 12B IT

12.0B parameters · 262,144 context

Dense Gemma4UnifiedForConditionalGeneration

Edge

Gemma 3n

Recent generation

Gemma 3n E2B

6.0B parameters · 32,768 context

Edge · Dense Gemma3nForConditionalGeneration

Gemma 4 Edge

Current generation 2 sibling variants

Gemma 4 E2B IT

5.1B parameters · 131,072 context

Edge · Dense Gemma4ForConditionalGeneration

Gemma 4 E4B IT

8.0B parameters · 131,072 context

Edge · Dense Gemma4ForConditionalGeneration

Deployment ecosystem

Choose a specific Gemma release and certified checkpoint before sizing hardware. FusionPointAI rejects unknown runtime configurations, calculates resident memory from total model weights, then distinguishes globally compatible GPUs from currently rentable inventory.

Validate the exact checkpoint license, runtime version, precision kernel, context length, concurrency and multi-GPU topology. Active MoE parameters affect compute, but resident model memory continues to use total stored weights.

Common use cases

On-device prototypes

Private assistants

Vision-language tasks

License overview

Licenses are certified per release, not inferred from the provider. Review the release page before commercial deployment because terms can differ between generations and variants.

Commercial-use status is a source-backed catalogue classification, not legal advice. Follow the linked official license and attribution requirements.

Release timeline

google/gemma/gemma-4-26b-a4b-nvfp4

3 Sept 2026

Source details

Gemma models: Versions, specs, GPU requirements & hosting

Overview and history

Architecture and model range

Generations, tracks and variants

General

gemma-4-26b-a4b

Gemma-4-26B-A4B-NVFP4

gemma-4-31b-it

Gemma-4-31B-IT-NVFP4

Gemma 3

Gemma 3 12B IT

Gemma 3 27B IT

contextpilot-e4b

ContextPilot-E4B

gemma-4

gemma-4-E4B-it-qat-q4_0-unquantized

gemma-4-E2B-it-qat-q4_0-unquantized

gemma-4-26B-A4B-it-qat-q4_0-unquantized

gemma-4-12B-it-qat-q4_0-unquantized

gemma-4-E2B-it-qat-mobile-transformers

gemma-4-31B-it-qat-q4_0-unquantized

gemma-4-E2B-it

gemma-4-E4B-it-qat-mobile-transformers

gemma-4-E4B-it

gemma-4-26B-A4B-it

gemma-4-12B-it

gemma-4-31B-it

Gemma 4 26B-A4B IT

Gemma 4 31B IT

Gemma 4 12B IT

Edge

Gemma 3n

Gemma 3n E2B

Gemma 4 Edge

Gemma 4 E2B IT

Gemma 4 E4B IT

Deployment ecosystem

Common use cases

License overview

Release timeline

google/gemma/gemma-4-26b-a4b-nvfp4

google/gemma/gemma-4-31b-it-nvfp4

google/gemma/gemma-4-31b-it-qat-q4-0-unquantized

google/gemma/gemma-4-26b-a4b-it-qat-q4-0-unquantized

google/gemma/gemma-4-12b-it-qat-q4-0-unquantized

google/gemma/gemma-4-e4b-it-qat-q4-0-unquantized

google/gemma/gemma-4-e2b-it-qat-q4-0-unquantized

google/gemma/gemma-4-e4b-it-qat-mobile-transformers

google/gemma/gemma-4-e2b-it-qat-mobile-transformers

google/gemma/contextpilot-e4b

google/gemma/gemma-3-27b-it

google/gemma/gemma-3-12b-it

google/gemma/gemma-3n-e2b

google/gemma/gemma-4-31b-it

google/gemma/gemma-4-26b-a4b-it

google/gemma/gemma-4-12b-it

google/gemma/gemma-4-e4b-it

google/gemma/gemma-4-e2b-it

google/gemma/gemma-gemma-4-31b-it

google/gemma/gemma-gemma-4-12b-it

google/gemma/gemma-gemma-4-e4b-it

google/gemma/gemma-gemma-4-26b-a4b-it

google/gemma/gemma-gemma-4-e2b-it

google/gemma-4-31B-it @ 2026-07-21T17:43:09Z

google/gemma-4-E4B-it @ 2026-07-21T17:40:18Z

google/gemma-4-26B-A4B-it @ 2026-07-21T13:47:21Z

google/gemma-4-12B-it @ 2026-07-21T04:39:02Z

google/gemma-4-E2B-it @ 2026-07-20T17:41:50Z

Frequently asked questions

Official references