Public preview · figures are illustrative fixture data, attributed to their source · data to May 2025 · read-only
Model · Gemma 3 · March 2025 Live · Hugging Face

Gemma 3 27B

A 27 billion-parameterHow many numbers the model learned during training — the usual rough measure of its size. dense model from Google DeepMind. Multimodal models up to 27B with 128K context.

Size
27.4Bdense
Memory to run
~20 GBsmallest, 8K ctx
Context
128Ktokens
License
RestrictedGemma Terms of Use
Updated
18 Sept 2026

Which version to use

Note: KV estimate ignores sliding-window layers and overstates long-context memory.

Where it runs

Best-fitting version on each reference system, at an 8K contextHow much text the model can consider at once, counted in tokens — roughly ¾ of a word each..

Runs well 11

Slowly (CPU or offload) 2

Too large 0

None of the reference systems.

Benchmarks

Each benchmarkA fixed set of questions every model is given, so their scores can be compared on the same task. is developer-reported; prompts and settings differ between labs. Bars are relative to the best open result.

BenchmarkGemma 3 27B IT
Benchmark GPQA DiamondAccuracy
42.4src
Benchmark IFEvalStrict prompt accuracy
90.4src
Benchmark MMLU-ProAccuracy
67.5src

developer reported

Variants & downloads

Each variant is a separate set of weights. Expand one for its downloads: quantizationStoring each of the model’s numbers with fewer bits, so the file is smaller and needs less memory. trades a little quality for a much smaller file, and the memory column adds the working memoryExtra space the model needs while it answers. It grows with the length of the conversation, on top of the file itself. a conversation needs on top.

Variant Gemma 3 27B ITvision · Google DeepMind · 3 downloads · Gemma Terms of UseRestricted
What it is
Understands images as well as text.
Publisher
Google DeepMind
License
Gemma Terms of Use · commercial use restricted
Released
12 Mar 2025

Contributions are not open yet, so there is nothing here from members.

QuantizationFormatBitsDownloadMemory @ 8KPublisher
Download BF16nativesafetensors1651.1 GiB~57.5 GiBGoogle DeepMindgoogle/gemma-3-27b-it
Download Q4_K_Mk quantgguf4.8915.6 GiB~20.6 GiBLM Studio Communitylmstudio-community/gemma-3-27b-it-GGUF
Download MLX 4-bitmlxmlx4.514.4 GiB~19.3 GiBMLX Communitymlx-community/gemma-3-27b-it-4bit
Architecture detailsLayers, attention and KV cache geometry
Parameters
27.43B
Active / token
All (dense)
Architecture
Dense
Layers
62
Attention heads
32
KV heads
16
Head dim
128
KV cache @ 8K (fp16)
3.88 GiB
Max context
131,072 tokens

Measured performance

Throughput on specific systems and runtimes, with the source of each measurement.

No performance measurements recorded yet.

Community results

Contributions are not open yet, so there is nothing here from members.

Reviews

Contributions are not open yet, so there is nothing here from members.

Lineage

How this model’s variants relate to each other and to other models.

VariantGemma 3 27B IT vision

Sources & history

Sources

  • Hugging Face Hub (live source, 1 record, 18 Sept 2026)
  • Mutinai illustrative fixtures (illustrative fixture, 1 record, 13 Sept 2026)

External identifiers

Field history

  • license = "gemma" Hugging Face Hub (current)

Timeline

  1. Gemma 3 released with vision and 128K context