Marketplace · Text & chat

gemma3:12b

Call gemma3:12b through one OpenAI-compatible API. It's served by 1 host on the Inpherio grid, each at a price it sets — your request goes to the cheapest healthy one, and fails over to the next if it's down.

From
£0.06
per 1M output tokens
£1 buys about 17 million tokens (≈ 13 million words)
Hosts
1
Loaded now
0

At a glance.

vision

  • Type: Text & chat
  • Hosts serving it now: 1 (0 with it loaded in memory)
  • Engines: ollama
  • Context window: up to 131,072 tokens
python · openai sdk
from openai import OpenAI
client = OpenAI(base_url="https://inpherio.co.uk/v1",
                api_key=INPHERIO_API_KEY)
reply = client.chat.completions.create(
    model="gemma3:12b",
    messages=[{"role": "user", "content": "Hello"}],
)

gemma3:12b in detail.

The current, most capable model that runs on a single GPU.

Published by
google
Parameters
12.2B
Download size
8.2 GB (as packaged for Ollama)
Quantisation
Q4_K_M
Format
gguf
Architecture
gemma3
Licence
Gemma terms · access is gated (accept the terms on Hugging Face)
Based on
google/gemma-3-12b-pt
Task
image text to text
Library
transformers
First published
1 Mar 2025
Last updated
21 Mar 2025
Downloads (30 days)
365,795 on Hugging Face
Likes
908

gemma3 image-text-to-text

The model card

Published by the model's authors on Ollama — shown as they wrote it, unedited by Inpherio.

This model requires Ollama 0.6 or later. Download Ollama

Gemma is a lightweight, family of models from Google built on Gemini technology. The Gemma 3 models are multimodal—processing text and images—and feature a 128K context window with support for over 140 languages. Available in 270M, 1B, 4B, 12B, and 27B parameter sizes, they excel in tasks like question answering, summarization, and reasoning, while their compact design allows deployment on resource-limited devices.

Models

Text

270M parameter model (32k context window)

ollama run gemma3:270m

1B parameter model (32k context window)

ollama run gemma3:1b

Multimodal (Vision)

4B parameter model (128k context window)

ollama run gemma3:4b

12B parameter model (128k context window)

ollama run gemma3:12b

27B parameter model (128k context window)

ollama run gemma3:27b

Quantization aware trained models (QAT)

The quantization aware trained Gemma 3 models preserves similar quality as half precision models (BF16) while maintaining a lower memory footprint (3x less compared to non-quantized models).

1B parameter model

ollama run gemma3:1b-it-qat

4B parameter model

ollama run gemma3:4b-it-qat

12B parameter model

ollama run gemma3:12b-it-qat

27B parameter model

ollama run gemma3:27b-it-qat

Evaluation

Benchmark Results

Gemma 3 270M

Benchmark n-shot Gemma 3 270m instruction tuned

HellaSwag 0-shot 37.7

PIQA 0-shot 66.2

ARC-c 0-shot 28.2

WinoGrande 0-shot 52.3

BIG-Bench Hard few-shot 26.7

IF Eval 0-shot 51.2

These models were evaluated against a large collection of different datasets and metrics to cover different aspects of text generation:

Reasoning, logic and code capabilities

Benchmark Metric Gemma 3 PT 1B Gemma 3 PT 4B Gemma 3 PT 12B Gemma 3 PT 27B

HellaSwag 10-shot 62.3 77.2 84.2 85.6

BoolQ 0-shot 63.2 72.3 78.8 82.4

PIQA 0-shot 73.8 79.6 81.8 83.3

SocialIQA 0-shot 48.9 51.9 53.4 54.9

TriviaQA 5-shot 39.8 65.8 78.2 85.5

Natural Questions 5-shot 9.48 20.0 31.4 36.1

ARC-c 25-shot 38.4 56.2 68.9 70.6

ARC-e 0-shot 73.0 82.4 88.3 89.0

WinoGrande 5-shot 58.2 64.7 74.3 78.8

BIG-Bench Hard

28.4 50.9 72.6 77.7

DROP 3-shot, F1 42.4 60.1 72.2 77.2

AGIEval 3-5-shot 22.2 42.1 57.4 66.2

MMLU 5-shot, top-1 26.5 59.6 74.5 78.6

MATH 4-shot – 24.2 43.3 50.0

GSM8K 5-shot, maj@1 1.36 38.4 71.0 82.6

GPQA

9.38 15.0 25.4 24.3

MMLU (Pro) 5-shot 11.2 23.7 40.8 43.9

MBPP 3-shot 9.80 46.0 60.4 65.6

HumanEval pass@1 6.10 36.0 45.7 48.8

MMLU (Pro COT) 5-shot 9.7 NaN NaN NaN

Multilingual capabilities

Benchmark Gemma 3 PT 1B Gemma 3 PT 4B Gemma 3 PT 12B Gemma 3 PT 27B

MGSM 2.04 34.7 64.3 74.3

Global-MMLU-Lite 24.9 57.0 69.4 75.7

Belebele 26.6 59.4 78.0 –

WMT24++ (ChrF) 36.7 48.4 53.9 55.7

FloRes 29.5 39.2 46.0 48.8

XL-Sum 4.82 8.55 12.2 14.9

XQuAD (all) 43.9 68.0 74.5 76.8

Multimodal capabilities

Benchmark Gemma 3 PT 4B Gemma 3 PT 12B Gemma 3 PT 27B

COCOcap 102 111 116

DocVQA (val) 72.8 82.3 85.6

InfoVQA (val) 44.1 54.8 59.4

MMMU (pt) 39.2 50.3 56.1

TextVQA (val) 58.9 66.5 68.6

RealWorldQA 45.5 52.2 53.9

ReMI 27.3 38.5 44.8

AI2D 63.2 75.2 79.0

ChartQA 45.4 60.9 63.8

ChartQA (augmented) 81.8 88.5 88.7

VQAv2 – – –

BLINK 38.0 35.9 39.6

OKVQA 51.0 58.7 60.2

TallyQA 42.5 51.8 54.3

SpatialSense VQA 50.9 60.0 59.4

CountBenchQA 26.1 17.8 68.0

Reference

Gemma Terms of Use

Sources: google/gemma-3-12b-it on Hugging Face · gemma3:12b on Ollama. Collected from public sources on 8 Nov 2026; the host serving it may run a different build or quantisation — the marketplace shows each host's own measurements.

Who's serving it.

Every host benchmarked this model on its own hardware before serving it. Prices are set by each host; health is scored by the platform from what it observes.

Hosts serving gemma3:12b right now
HostTierPriceMeasured speedHealthStateRent
Lumaram-MacBookPro-M4Pro-002L1£0.03 in · £0.06 out26 tok/s · 116 ms first token80loads on request—

Live data, refreshed on every visit. How billing works.

Room on the grid for your machine.

Right now one machine serves gemma3:12b. If it drops, so does every request for it — a second host is what turns it into failover, and the grid needs that more than another copy. The router weighs price, measured speed, health and whether the model is already loaded — so a host with a sharper price or quicker hardware wins the next call.

Hosts now
1
0 loaded in memory
Price to beat
£0.06
per 1M output tokens
Fastest measured
26 tok/s

Set your own price — you keep 85% of what it earns, and new hosts start with 12 months fee-free (up to £250 of earnings). A payout fee of £2 + 1% per payout to a UK bank account, £2 + 2% to any other country applies when you're paid. Install the connector, load gemma3:12b, and the platform benchmarks it on your hardware before your first request.