Gemma 4 31B IT
Sovereignty and hosting
This model is catalogued for reference. It is not hosted on LLM EU yet, and no first-party endpoint serves it here. Requests to it go to the provider directly, under that provider's own terms.
Summary
Gemma 4 31B IT is the largest instruction-tuned model of Google DeepMind’s Gemma 4 generation, published under Apache-2.0 with a 256K-token context window. It processes text and images with variable resolution and generates text; the vendor states multilingual support in over 140 languages. The model card reports 85.2% on MMLU Pro, 84.3% on GPQA Diamond and 88.4% on MMMLU, all self-reported.
Specifications
- Provider
- Google DeepMind (US)
- Context window
- 262k
- Parameters
- 31.27B
- Modalities
- text, vision
- License
- apache-2.0 · licence
- Open weights
- yes
- Release date
- 11 Mar 2026
- Tasks
- vision, reasoning, multilingual-support, coding, summarization
Languages
Strong: en, de, fr, es, it, pt
Adequate: nl, pl, sv, cs, ro, el, uk, ru, ar, hi, zh, ja, ko, tr
Languages
Strong
- en
- de
- fr
- es
- it
- pt
Adequate
- nl
- pl
- sv
- cs
- ro
- el
- uk
- ru
- ar
- hi
- zh
- ja
- ko
- tr
Not verified: the provider has not documented this claim. — a language is listed as strong only where a source claims it; the card links the evidence where one exists.
When to use it
Strengths
- 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
Limits
- 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