GoogleUnited States· released Feb 2026

Gemini 3.1 Pro

Name: Gemini 3.1 Pro
Author: Google

Ranked #15, 26.6% behind the frontier on average.

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

Third-party publishedHigh confidence (12+ comparable tests, all 4 Index categories)

Available in Theo as Theo Insight

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

73.4

Ranked #15 (11 to 17) of 36 · 26.6% behind the frontier on average · 22 comparable tests · 4 counting categories

Drop any single test and the Index lands between 69.9 and 75.7, anywhere from #11 to #17. Neighbours inside that range are ties.

7 of 8 hardest tests · share of the frontier

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

What it's known for

Gemini 3.1 Pro, in 22 results.

Strongest in reasoning (88.4) and agentic (80.5). It trails the frontier most in math, 46.8% behind on average. No comparable result yet in long context. Between 25% and 30% behind the best published results, on average. Drop any single test and it would sit anywhere from #11 to #17.

Strongest category

Reasoning

Hard, novel problems: graduate science, abstract puzzles, expert exams.

88.4

Humanity's Last Exam46.4% · 100% of best
ARC-AGI-198.0% · 100% of best
GPQA Diamond94.4% · 98% of best
SimpleBench79.6% · 97% of best
ARC-AGI-277.1% · 81% of best
CritPt17.7% · 55% of best

Mean of 6 of the category's 6 comparable tests. Counts toward the Index.

Best single result

Humanity's Last Exam

Answer an expert-written question from the frontier of a hundred academic fields, with an exact answer and a stated confidence.

46.4%

99.9% of the best (46.5%) · #2 of 18.

The hardest general exam in the set. It separates models that reason from models that remember.

Where it trails

Code Arena (WebDev)

Build a web app from the same prompt as a rival model; real users vote blind on the result.

1,446

76.6 points short of parity with the board leader (1,797) · #44 of 86.

People judging finished apps side by side is the most honest measure of front-end quality.

Not measured yet

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

Long context

Covered, but not averaged into the Index:

Knowledgesingle-test category, shown beside the Index · 97.2
Multimodalsingle-test category, shown beside the Index · 90.1
Human preferencesingle-test category, shown beside the Index · 94.1

Every score, with its source

See it at work

What Gemini 3.1 Pro 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.

Knowledge · mean 97.2 across 1 comparable test · a single-test category, shown beside the Index and never averaged into it

SimpleQA VerifiedEpoch-run

Answer a short factual question exactly, without searching, or say you do not know.

73.5%

97.2% of the best (75.6%)
#2 of 57 · n = 1,000

preview · high

Source: Epoch AI (internal runs)

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

Gemini 3.1 Pro 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.

Gemini 3.1 Pro

#15·Index 73.4B-

1 : 10

wins · 2 ties · 13 shared

GPT-6 Astra

#1·Index 99.0A+

GPT-6 Astra comes out ahead on 10 of the 13 tests both have taken.

Category means

Reasoning
88.4
99.6
Knowledge
97.2
100.0
Coding
71.6
97.7
Math
53.2
99.8
Agentic
80.5
98.5
Multimodal
90.1
—
Human preference
94.1
—

Shared tests · biggest gaps first

Code Arena (WebDev)Coding
1,446
1,797
−76.6
ProofBenchMath
26.0%
99.0%
−73.0
FrontierMath Tier 4Math
26.8%
97.6%
−72.5
CritPtReasoning
17.7%
31.7%
−43.4
FrontierMath (Tiers 1–3)Math
59.6%
93.7%
−36.3
APEX-AgentsAgentic
33.5%
46.7%
−27.8
WeirdMLCoding
72.1%
92.9%
−22.4
ARC-AGI-2Reasoning
77.1%
95.0%
−18.8
Mock AIME 2024–2025Math
95.6%
100.0%
−4.4
SciCodeCoding
58.9%
56.5%
+4.0
SimpleQA VerifiedKnowledge
73.5%
75.6%
−2.8
GPQA DiamondReasoning
94.4%
95.8%
tie
ARC-AGI-1Reasoning
98.0%
98.5%
tie

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

Reasoning88.4 · 6/6
Knowledgebeside97.2 · 1/1
Coding71.6 · 4/7
Math53.2 · 4/5
Agentic80.5 · 5/8
Multimodalbeside90.1 · 1/1
Human preferencebeside94.1 · 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

94.4%

98.1% of the best (95.8%), above chance
96th percentile of 78 · n = 198

preview · high

Gemini 3.1 Pro

Gemini 3.1 Pro, in 22 results.

What Gemini 3.1 Pro was asked to do, and how it did.

Gemini 3.1 Pro against whoever you pick.

Every test, every source

Compare with its neighbours.

The best models, ranked here, working inside Theo.