GoogleUnited States· released Apr 2026· open weights

Gemma 4 31B

Name: Gemma 4 31B
Author: Google

Provisional: 9 comparable tests so far. Ranked once it reaches 6 across 3 counting categories.

Google, United States. One of 14 models from this lab on the Index.

Third-party publishedLow confidence (thin coverage)

Available in Theo as Theo Open Lite

Theo orchestrates it for the steps it does best and always shows which engine answered.

Use it in Theo

Snapshot September 8, 2026

OpenCharts Index

57.0

Provisional: not enough comparable tests or counting categories to rank yet · 9 comparable tests · 0 counting categories

Drop any single test and the Index lands between 51.3 and 65.6.

Needs more results (0 of 8 so far)

Distance to today's frontier on the hardest tests. Not a measure of intelligence.

What it's known for

Gemma 4 31B, in 9 results.

Its widest gap is on CritPt, 95.6% behind the best published result. No comparable results yet in agentic and long context. Between 40% and 50% behind the best published results, on average.

Strongest category

No category counts yet: a category counts once it has more than one comparable test and the model has taken at least half of them. The scores it has are below.

Best single result

Vision Arena

Answer prompts that include an image; real users vote blind between two models' answers.

1,261

43% expected win rate against the board leader (1,313) · #27 of 54.

Seeing is half of most real tasks: screenshots, documents, photos and charts.

Where it trails

CritPt

Solve an unpublished research-level physics problem to a numeric or symbolic answer checked by an official server.

1.4%

95.6% behind the best published result (32.3%) · #59 of 78.

Research physics is far past textbook recall. Only a few models produce anything a physicist would accept.

Not measured yet

No comparable result yet in these 2 categories. A blank is a blank, never a zero, and it does not lower the Index.

Agentic
Long context

Covered, but not averaged into the Index:

Reasoning2 of 6 tests · needs 3 · mean 38.1
Knowledgesingle-test category, shown beside the Index · 13.8
Coding3 of 7 tests · needs 4 · mean 47.2
Math1 of 5 tests · needs 3 · mean 73.3
Multimodalsingle-test category, shown beside the Index · 85.2
Human preferencesingle-test category, shown beside the Index · 84.1

Every score, with its source

See it at work

What Gemma 4 31B was asked to do, and how it did.

Every test it has taken, by category, with the kind of task it faced, who produced the number and where the result landed against the best published one. The examples are original and illustrative, never items from the datasets themselves.

Multimodal · mean 85.2 across 1 comparable test · a single-test category, shown beside the Index and never averaged into it

Vision ArenaLeaderboard

Answer prompts that include an image; real users vote blind between two models' answers.

1,261

43% expected win rate against the board leader (1,313)
#27 of 54

board 2026-08-27

Source: Arena

Every result is read against the best published one on its test: percent scores as a share of the best above chance, Arena ratings as the expected win rate against the board leader, open-ended values on a log scale. Sitting a harder exam never lowers a model. Every number links to the publisher that produced it.

Head to head

Gemma 4 31B against whoever you pick.

Choose a rival. Every counted test both have taken appears side by side as a share of the best published result on that test, with the category means above. A win is a higher share; a gap under 2 points is a tie, because one item on a 45-problem exam is 2.2 points and publishers' own standard errors are of that order.

Gemma 4 31B

provisional·Index 57.0C-

0 : 7

wins · 0 ties · 7 shared

GPT-6 Astra

#1·Index 99.0A+

GPT-6 Astra comes out ahead on 7 of the 7 tests both have taken.

Category means

Reasoning
38.1
99.6
Knowledge
13.8
100.0
Coding
47.2
97.7
Math
73.3
99.8
Agentic
—
98.5
Multimodal
85.2
—
Human preference
84.1
—

Shared tests · biggest gaps first

CritPtReasoning
1.4%
31.7%
−93.8
SimpleQA VerifiedKnowledge
10.4%
75.6%
−86.2
Code Arena (WebDev)Coding
1,363
1,797
−84.8
WeirdMLCoding
52.3%
92.9%
−43.7
GPQA DiamondReasoning
75.8%
95.8%
−28.3
Mock AIME 2024–2025Math
73.3%
100.0%
−26.7
SciCodeCoding
43.4%
56.5%
−21.0

Bars are distances to the best published result on each test (percent scores above chance, Arena ratings as win rate against the leader, open-ended values on a log scale). Raw values as the publishers report them.

Open GPT-6 Astra

Scores

Every test, every source

The best published run per benchmark, the raw value as the publisher reports it, who produced the number, how close it comes to the best published result on that test, and where the model sits among every model scored on it. Percent scores are read above chance, Arena ratings as a win rate against the board leader, open-ended values on a log scale. A test with too few models to compare against is shown but not counted.

Category means

Reasoning38.1 · 2/6
Knowledgebeside13.8 · 1/1
Coding47.2 · 3/7
Math73.3 · 1/5
Agentic—
Multimodalbeside85.2 · 1/1
Human preferencebeside84.1 · 1/1
Long context—

Mean · tests taken of the category's comparable tests. A faded category either rests on fewer than half of them or is a single-test category (marked “beside”), which is shown next to the Index and never averaged into it.

GPQA DiamondEpoch-runReasoning

75.8%

71.7% of the best (95.8%), above chance
8th percentile of 78 · n = 198

minimal

Source: Epoch AI — AI Benchmarking Hub

CritPtLeaderboardReasoning

1.4%

4.4% of the best (32.3%)
23th percentile of 78

Source: Epoch AI — AI Benchmarking Hub

SimpleQA VerifiedEpoch-runKnowledge

10.4%

13.8% of the best (75.6%)
0th percentile of 57 · n = 1,000