ITG NowHome › AI models › Gemma 4 31B
Gemma 4 31B
Google · Released 31 Mar 2026 · Open weights
Google's open model you can run yourself; best for simple tasks.
Key facts
- Maker
- Released
- 31 Mar 2026
- Open weights
- Yes, you can download it and run it yourself
- License
- Apache 2.0: free for almost any use
- Memory (context window)
- about 262 thousand tokens (how much it can read at once)262,144 tokens exactly
- Size
- 31.3B parametersB = billion, T = trillion. Bigger usually needs a stronger computer.
- Official page
- Official page for Gemma 4 31B (ai.google.dev)
Scores
"Independent" = run by an outside group. "Maker's own number" = the company's own claim.
All-roundArtificial Analysis Intelligence Index
15 index points #16 of 17 models scored independent
One combined score from many independent tests of how smart a model is.
as of 17 Oct 2026 · source
Chat & writingLMArena Text
1452 Arena score #10 of 15 models scored independent
When real people compare two AI answers side by side without knowing which AI wrote them, which one do they pick more often?
as of 15 Oct 2026 · source
Build websites & appsLMArena WebDev
1367 Arena score #16 of 18 models scored independent
When people ask two AIs to build the same website or app and compare the results, which one do they pick?
as of 17 Oct 2026 · source
CodingLiveBench 2.0 Coding
49.0 score out of 100 #15 of 15 models scored independent
Fresh coding tasks with tests that check the answer, changed often so models cannot memorize them.
setting: highest supported effort · as of 14 Oct 2026 · source
ReasoningLiveBench 2.0 Reasoning
68.3 score out of 100 #15 of 15 models scored independent
Fresh logic and puzzle questions with one right answer, changed often so models cannot memorize them.
setting: highest supported effort · as of 14 Oct 2026 · source
Science questionsGPQA Diamond
84.3% #6 of 6 models scored maker's own number
Very hard science questions in biology, chemistry and physics, written by PhD experts.
as of 17 Oct 2026 · source
Hardest examHumanity's Last Exam (no tools)
19.5% #6 of 6 models scored maker's own number
2,500 expert questions written to be too hard for AI, answered without web search or code.
as of 17 Oct 2026 · source
SpeedOutput speed
35 tokens per second #16 of 16 models scored independent
How many tokens (pieces of words) per second the model writes once it starts answering. Higher is faster.
as of 17 Oct 2026 · source
How it compares
The 3 models closest to Gemma 4 31B on the Artificial Analysis Intelligence Index, an all-round score. Gemma 4 31B: 15 index points, #16 of 17 models scored.
- Mistral AIMistral Medium 3.5Open weights. License: Modified MIT: free to use with conditions14 index points#17 of 17 models scored
- GoogleGemini 3.5 Flash-Lite22 index points#15 of 17 models scored
- Reflection AIBeam 1.0Open weights. License: Apache 2.0: free for almost any use30 index points#14 of 17 models scored
News about Gemma 4 31B on ITG Now
No stories about Gemma 4 31B on ITG Now yet.
Related free tools
- Artemis: Google's Free AI Android Automation ToolArtemis is free AI Android automation from Google. Type a task in plain English and an AI taps, swipes and tests apps on a real Android phone for you.
FAQ
Is Gemma 4 31B free and open?
Gemma 4 31B has open weights: you can download it and run it on your own computer or server. License: Apache 2.0: free for almost any use.
How much does Gemma 4 31B cost?
There is no listed price per token. It has open weights, so you can run it yourself or through a host you pick.
What is Gemma 4 31B best at?
Google's open model you can run yourself; best for simple tasks. Its best result: #6 of 6 models scored on GPQA Diamond (science questions).
Who makes Gemma 4 31B?
Gemma 4 31B is made by Google. It came out on 31 Mar 2026.