Model

Gemini 3.1 Pro

CLOSED

Google · Gemini 3.1 family · released Feb 19, 2026

Coding-arena leader with extended long-context frontier reasoning; ARC-AGI-2 77.1%, 1M-token context.

ReasoningCodingVisionFunction callingTool useAgentic

2538.5

Elo · rank #35

Parameters

Undisclosed

Active params

Undisclosed

Context

1M tokens

Architecture

Sparse Mixture-of-Experts Transformer (configuration undisclosed)

License

Proprietary

Languages

—

API price (in/out)

$2 / $12

Modalities

text · vision · audio · video

Benchmark results

Bar shows position within the tracked field; marker = field best

Agents' Last ExamAgents32.0%#16

best: GPT-6 Astra · 59.3%

ARC-AGI-2Reasoning77.1%#11

best: GPT-6 Astra · 95.0%

Arena EloHuman preference1487#11

best: Claude Fable 5 · 1504

BrowseCompAgents85.9%#8

best: Kimi K3 · 91.2%

GDPval-AAAgents1317#31

best: Claude Fable 5 · 1932

GPQA DiamondReasoning94.3%#4

best: GPT-6 Astra · 96.0%

Humanity's Last ExamReasoning44.4%#20

best: Claude Opus 5.5 · 67.7%

LiveCodeBench ProCoding2887#1

best: this model · 2887

LiveCodeBenchCoding88.5%#14

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MMMLUKnowledge92.6%#1

best: this model · 92.6%

MMMU-ProVision80.5%#11

best: Claude Opus 4.7 · 85.5%

SWE-bench ProCoding54.2%#41

best: Claude Opus 5.5 · 89.9%

SWE-bench VerifiedCoding78.8%#32

best: Claude Opus 5 · 97.0%

τ²-Bench RetailAgents90.8%#2

best: Claude Sonnet 4.6 · 91.7%

τ²-Bench TelecomAgents99.3%#2

best: Claude Opus 4.6 · 99.3%

Terminal-Bench 2.0Coding68.5%#34

best: DeepSeek-V4.1-Flash · 90.6%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

Output / M tok

$12

Gemini 3.1 family

Elo progression across releases

Gemini 3.1 ProFeb 20262538.5 Gemini 3.1 Flash-LiteMar 20261852.1

Resources

DOCShttps://deepmind.google/models/gemini/pro/APIAvailable via Google API

API price $2/$12 · each benchmark row carries its own source badge (see methodology)