GPT-5.4 ProvsGemini 3.1 Pro Preview

Across 6 shared benchmarks, GPT-5.4 Pro leads overall: GPT-5.4 Pro wins 6, Gemini 3.1 Pro Preview wins 0, with 0 ties and an average score difference of +8.48.

OpenAI
GPT-5.4 Pro

OpenAI · 2026-03-05 · Multimodal model

Google Deep Mind
Gemini 3.1 Pro Preview

Google Deep Mind · 2026-02-20 · Multimodal model

GPT-5.4 Pro6 wins(100%)(0%)0 winsGemini 3.1 Pro Preview

Benchmark scores

Grouped by capability, sorted by largest gap within each. 6 shared benchmarks.

General Knowledge

GPT-5.4 Pro 3/3
BenchmarkGPT-5.4 ProGemini 3.1 Pro PreviewDiff
HLE58.703 / 157Thinking High (With Tools)51.4015 / 157Thinking High (With Tools)+7.30
ARC-AGI-283.306 / 59Thinking High (No Tools)77.107 / 59Thinking High (No Tools)+6.20
GPQA Diamond94.402 / 178Thinking High (No Tools)94.303 / 178Thinking High (No Tools)+0.10

Math and Reasoning

GPT-5.4 Pro 2/2
BenchmarkGPT-5.4 ProGemini 3.1 Pro PreviewDiff
FrontierMath - Tier 437.505 / 80Normal (With Tools + Internet)16.7020 / 80Normal (No Tools)+20.80
FrontierMath503 / 60Thinking High (No Tools)36.9011 / 60Thinking High (No Tools)+13.10

AI Agent - Information Search

GPT-5.4 Pro 1/1
BenchmarkGPT-5.4 ProGemini 3.1 Pro PreviewDiff
BrowseComp89.302 / 45Thinking High (With Tools)85.903 / 45Thinking High (With Tools + Internet)+3.40

Specs

FieldGPT-5.4 ProGemini 3.1 Pro Preview
PublisherOpenAIGoogle Deep Mind
Release date2026-03-052026-02-20
Model typeMultimodal modelMultimodal model
ArchitectureDenseDense
ParametersNot availableNot available
Context length1M1M
Max output125K32K

API pricing

Prices use DataLearner records when available; missing fields are not inferred.

ItemGPT-5.4 ProGemini 3.1 Pro Preview
Text input$30 / 1M tokens$2 / 1M tokens
Text output$180 / 1M tokens$12 / 1M tokens

Summary

  • GPT-5.4 Proleads in:General Knowledge (3/3), Math and Reasoning (2/2), AI Agent - Information Search (1/1)

On average across the 6 shared benchmarks, GPT-5.4 Pro scores 8.48 higher.

Largest single-benchmark gap: FrontierMath - Tier 4 — GPT-5.4 Pro 37.50 vs Gemini 3.1 Pro Preview 16.70 (+20.80).

Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.