Google DeepMind
Gemma 4 31B
Part of Google DeepMind’s Gemma 4 family. Registry facts are limited to what Google DeepMind states on its own pages, each linked to where we found it. Benchmark results carry their own source and provenance label.
CurrentVerified against official sources
Verified facts
“source” links to the official page each value was confirmed on
- Organization
- Google DeepMind
- Family
- Gemma 4
- Release date
- (released)source for Release date (external site)
- Context window
- 262,144 tokensStated as “256K tokens (model card)”source for Context window (external site)
- Max output tokens
- Not verified
- Knowledge cutoff
- January 2025 (pre-training data cutoff, per model card)source for Knowledge cutoff (external site)
- Input modalities
- Text, Imagesource for Input modalities (external site)
- Output modalities
- Textsource for Output modalities (external site)
- Weights
- Open weightssource for Weights (external site)
- API model ID
gemma-4-31b-itsource for API model ID (external site)- Status
- Current
- Last verified
Reported benchmark results
5 results on file, all reported by Google DeepMind. Scores are copied exactly as published, with the settings the source stated. Rows are different measurements — compare a score only with others in the same benchmark variant, via the variant link.
| Benchmark | Variant | Model detail | Score | Settings (as stated) | Source | Evidence |
|---|---|---|---|---|---|---|
| AIME | 2026 — see all AIME results for this variant | Not stated by source | 89.2%No CI reported | no tools; instruction-tuned model | Show excerpt for Gemma 4 31B on AIME, 2026“| AIME 2026 no tools | 89.2% |” Retrieved | |
| GPQA Diamond | No variant stated — see all GPQA Diamond results for this variant | Not stated by source | 84.3%No CI reported | instruction-tuned model; no other conditions stated | Vendor-reportedGoogle DeepMindhuggingface.co — source for Gemma 4 31B on GPQA Diamond, No variant stated (external site) | Show excerpt for Gemma 4 31B on GPQA Diamond, No variant stated“| GPQA Diamond | 84.3% |” Retrieved |
| HLE | No tools — see all HLE results for this variant | Not stated by source | 19.5%No CI reported | no tools; instruction-tuned model | Vendor-reportedGoogle DeepMindhuggingface.co — source for Gemma 4 31B on HLE, No tools (external site) | Show excerpt for Gemma 4 31B on HLE, No tools“| HLE no tools | 19.5% |” Retrieved |
| LiveCodeBench | v6 — see all LiveCodeBench results for this variant | Not stated by source | 80.0%No CI reported | instruction-tuned model; no other conditions stated | Vendor-reportedGoogle DeepMindhuggingface.co — source for Gemma 4 31B on LiveCodeBench, v6 (external site) | Show excerpt for Gemma 4 31B on LiveCodeBench, v6“| LiveCodeBench v6 | 80.0% |” Retrieved |
| MMLU-Pro | No variant stated — see all MMLU-Pro results for this variant | Not stated by source | 85.2%No CI reported | instruction-tuned model; no other conditions stated | Vendor-reportedGoogle DeepMindhuggingface.co — source for Gemma 4 31B on MMLU-Pro, No variant stated (external site) | Show excerpt for Gemma 4 31B on MMLU-Pro, No variant stated“| MMLU Pro | 85.2% |” Retrieved |
- Vendor-reported:
- Published by the organisation that built the model. Not independently reproduced.