Gemma 4 31B vs Gemma 3 27B
Gemma 4 31B is the newer model in the same family as Gemma 3 27B. Here is what changed. The best value in each row is highlighted, and every number links to its source.
- Gemma 4 31B leads Gemma 3 27B by 85 points on the LMArena text leaderboard, so people prefer its answers in about 62% of head-to-head votes.
- Gemma 4 31B is the newer model, released 13 months after Gemma 3 27B.
| Fact | GoogleGemma 4 31B | GoogleGemma 3 27B |
|---|---|---|
| Released | Apr 2026Google docs, Oct 1, 2026 | Mar 2025 |
| LMArena overall | 1443 (best)LMArena text leaderboard, Nov 11, 2026 | 1358LMArena text leaderboard, Nov 11, 2026 |
| LMArena coding | 1459 (best)LMArena text leaderboard, Nov 11, 2026 | 1322LMArena text leaderboard, Nov 11, 2026 |
| Context window | 256K tokensGoogle docs, Nov 17, 2026 | Not published |
| Max output | 32K tokensGoogle docs, Nov 17, 2026 | Not published |
| Reasoning mode | YesGoogle docs, Nov 17, 2026 | Not published |
| Takes images | YesGoogle docs, Nov 17, 2026 | Not published |
| Open weights | YesHugging Face: Gemma 4 31B, Oct 1, 2026 | YesHugging Face: Gemma 3 27B, Oct 1, 2026 |
LMArena scores come from anonymous head-to-head votes. Two scores only count as different when their margins of error don't overlap.
Compare Gemma 4 31B or Gemma 3 27B with others
More matchups
- ChatGPT vs Claude
- ChatGPT vs Gemini
- Claude vs Gemini
- ChatGPT vs Grok
- Gemini vs Grok
- Claude vs Grok
- ChatGPT vs DeepSeek
- Claude vs DeepSeek
- Gemini vs DeepSeek
- DeepSeek vs Qwen
- DeepSeek vs Kimi
- Qwen vs Kimi
- Meta AI vs Qwen
- Mistral vs Meta AI
Build your own comparison or see the best AI models by price, context and score.
Think you know these models? Test it in today’s quiz.
Play the daily