Title: Gemma 4 12B It vs Gemma 4 31B IT: Parameters, Licentie %E2%80%94 LLM EU
Canonical: https://llmeu.com/nl/compare/gemma-4-12b-it-vs-gemma-4-31b-it/
Markdown mirror: https://llmeu.com/nl/compare/gemma-4-12b-it-vs-gemma-4-31b-it.md
Machine-readable: https://llmeu.com/llms.txt (index) - https://llmeu.com/llms-full.txt (full catalogue) - https://llmeu.com/llms-catalogue.json

# Gemma 4 12B It vs Gemma 4 31B IT

Zet twee tot vier modellen naast elkaar op de velden die een deployment bepalen: context, prijs, regio, licentie, talen, hosting en wat is geverifieerd.

Waar twee modellen verschillen, toont deze tabel het verschil en stopt. LLM EU publiceert geen winnaarsbadge zonder een gedocumenteerde score voor die taak, en een ontbrekend cijfer blijft ontbreken in plaats van een nul te worden.

## Wanneer te gebruiken

### [Gemma 4 12B It](/nl/models/gemma-4-12b-it/)

#### Sterke punten

- open weights are published on Hugging Face and can be downloaded without a vendor agreement
- Apache-2.0 licence, so commercial self-hosting is permitted without a separate agreement
- at 12B parameters it fits on a single accelerator
- instruction-tuned checkpoint with a chat template in the repository
- speech recognition runs locally, so audio never leaves the operator premises

#### Beperkingen

- catalogue only %E2%80%94 LLM EU does not run this model, and nothing on this card implies an LLM EU endpoint
- no evaluation has been run by LLM EU
- the published configuration records no context window, so the card carries none
- at 12B parameters, self-hosting needs a single accelerator with enough memory for the weights
- vision checkpoints are heavier to serve than their parameter count alone suggests, because the image encoder and pre-processing run alongside the language model

#### Taken

[Meertalige ondersteuning](/nl/tasks/multilingual-support/), [Visie](/nl/tasks/vision/), [Samenvatten](/nl/tasks/summarization/)

### [Gemma 4 31B IT](/nl/models/gemma-4-31b-it/)

#### Sterke punten

- Apache-2.0 licence, unlike earlier Gemma generations
- 256K-token context window
- text and image input with variable resolution
- vendor reports multilingual coverage in over 140 languages

#### Beperkingen

- all published scores are self-reported by the vendor
- 31B parameters need a large-memory accelerator at BF16
- no evaluation has been run by LLM EU

#### Taken

[Visie](/nl/tasks/vision/), [Redeneren](/nl/tasks/reasoning/), [Meertalige ondersteuning](/nl/tasks/multilingual-support/), [Coderen](/nl/tasks/coding/), [Samenvatten](/nl/tasks/summarization/)

**Taken:** [Meertalige ondersteuning](/nl/tasks/multilingual-support/), [Visie](/nl/tasks/vision/), [Samenvatten](/nl/tasks/summarization/)

## Modellen vergelijken

- [Aya 23 8B vs Gemma 4 12B It](/nl/compare/aya-23-8b-vs-gemma-4-12b-it/)
- [Aya Expanse 8B vs Gemma 4 12B It](/nl/compare/aya-expanse-8b-vs-gemma-4-12b-it/)
- [Command R (08-2024) vs Gemma 4 31B IT](/nl/compare/command-r-08-2024-vs-gemma-4-31b-it/)
- [DeepSeek Coder V2 Lite Instruct vs Gemma 4 31B IT](/nl/compare/deepseek-coder-v2-lite-instruct-vs-gemma-4-31b-it/)
- [DeepSeek-R1-Distill-Qwen-32B vs Gemma 4 31B IT](/nl/compare/deepseek-r1-distill-qwen-32b-vs-gemma-4-31b-it/)
- [DeepSeek V2 Lite vs Gemma 4 31B IT](/nl/compare/deepseek-v2-lite-vs-gemma-4-31b-it/)