Gemma 4 31B
Google · Fast & cheap · Mar 31, 2026Open weights
The largest of Google's open-weight Gemma 4 family, small enough to host yourself and released under Apache 2.0.
- Artificial Analysis Intelligence Index
- 14.7#27 of 29
- API price per 1M tokens
- Self-hosted
- Context window
- 256Ktokens
- Output speed
- 35 t/s51s to first answer
Benchmark results
Each row shows this model (blue) against every other model in the catalogue (grey), worst on the left, best on the right.
Overall capability
- Artificial Analysis Intelligence Index14.7#27 of 29
Coding
- Arena WebDev1,365#25 of 27
Multimodal
- Arena Vision1,262#11 of 15
Human preference
- Arena Text1,453#18 of 24
Reasoning effort trade-offs
The same model at different thinking budgets. More effort buys a higher intelligence score at a higher cost per task and a longer wait. Measured by Artificial Analysis.
| Effort | Score | Cost / task | First answer | Speed |
|---|---|---|---|---|
| No reasoning | 13.9 | Not measured | 2s | 43 t/s |
| Default (headline score) | 14.7 | $0.000 | 51s | 35 t/s |
Specs & pricing
- Developer
- Released
- Mar 31, 2026
- Availability
- Generally available
- Weights
- Open · Apache 2.0
- Parameters
- 30.7B
- Reasoning
- Adjustable effort
- Knowledge cutoff
- Jan 2025
- Accepts
- Text, Image
- Produces
- Text
- Context window
- 256,000 tokens
- API price
- Self-hosted
Closest alternatives
Models from other labs nearest to Gemma 4 31B on Artificial Analysis Intelligence Index.
Mistral Medium 3.5
Mistral AI
Intelligence 14.2 · $1.50 / $7.50
Compare side by sideMuse Glimmer 30B
Meta
Intelligence 17.5 · Self-hosted
Compare side by sideMistral Small 4
Mistral AI
Intelligence 11.3 · $0.15 / $0.60
Compare side by side