Gemini 3.1 Pro vs GPT-5.4 nano
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
Gemini 3.1 Pro is the stronger all-rounder.GPT-5.4 nano is cheaper.
Scores updated · 46 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking50Gemini 3.1 Pro5 of 5 tests
- CodingWriting and fixing software51Gemini 3.1 Pro5 of 6 tests
- AgentsCarrying out multi-step tasks on its own52Gemini 3.1 Pro5 of 7 tests
- FactsGetting facts right instead of making them up31Gemini 3.1 Pro3 of 4 tests
- MathCompetition and research-level math20Gemini 3.1 Pro2 of 3 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video20Gemini 3.1 Pro2 of 2 tests
- Following instructionsDoing exactly what it is asked20Gemini 3.1 Pro2 of 2 tests
- Long documentsFinding answers in very long texts10Gemini 3.1 Pro1 of 1 test
Long documents rests on a single test.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
GPT-5.4 nano costs 90% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Gemini 3.1 Pro pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+71.4points ahead
- Short factual questions, answered correctlySimpleQA Verified+61.8points ahead
- Completes tasks by operating a computer desktopOSWorld-Verified+37.2points ahead
Where GPT-5.4 nano pulls ahead
- Real work tasks from 44 professionsGDPVal+7.8points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+6points ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+2.7points ahead
Every test, side by side
All 46 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemini 3.1 Pro
- Terminal-Bench 2.1Gemini 3.1 Pro by 13.173.860.7+13.1
- SciCodeGemini 3.1 Pro by 11.558.747.2+11.5
- Terminal-Bench HardGemini 3.1 Pro by 11.453.842.4+11.4
- LiveBench · CodingGemini 3.1 Pro by 5.776.570.8+5.7
- LiveBench · Agentic CodingGPT-5.4 nano by 2.744.146.8+2.7
- SWE-bench ProGemini 3.1 Pro by 1.854.252.4+1.8
AgentsGemini 3.1 Pro
- OSWorld-VerifiedGemini 3.1 Pro by 37.276.239+37.2
- MCP AtlasGemini 3.1 Pro by 13.169.256.1+13.1
- GDPValGPT-5.4 nano by 7.814.722.5+7.8
- AA ApexAgentsGemini 3.1 Pro by 7.13224.9+7.1
- τ-Bench V3 · BankingGPT-5.4 nano by 621.427.4+6
- AA IT-Bench SREGemini 3.1 Pro by 5.930.324.4+5.9
- Terminal-Bench 4.0Gemini 3.1 Pro by 3.540.5+3.5
ReasoningGemini 3.1 Pro
- ARC-AGI-2Gemini 3.1 Pro by 71.477.15.7+71.4
- Humanity's Last ExamGemini 3.1 Pro by 18.74728.3+18.7
- GPQA DiamondGemini 3.1 Pro by 12.494.181.7+12.4
- CritPtGemini 3.1 Pro by 8.417.79.3+8.4
- LiveBench · ReasoningGemini 3.1 Pro by 2.98481.1+2.9
FactsGemini 3.1 Pro
- SimpleQA VerifiedGemini 3.1 Pro by 61.873.511.7+61.8
- AA-Omniscience · AccuracyGemini 3.1 Pro by 29.254.925.7+29.2
- AA-Omniscience · Non-hallucinationGemini 3.1 Pro by 23.349.125.8+23.3
- Vectara HHEM hallucination ratelower is betterGPT-5.4 nano by 7.310.43.1+7.3
Images and chartsGemini 3.1 Pro
- MMMU-ProGemini 3.1 Pro by 1782.465.4+17
- LMArena · VisionGemini 3.1 Pro by 100 rating points12961196+100 rating
MathGemini 3.1 Pro
- FrontierMath Tiers 1-3 (v2)Gemini 3.1 Pro by 14.859.644.9+14.8
- FrontierMath Tier 4Gemini 3.1 Pro by 14.626.812.2+14.6
- LiveBench · Mathematicstie9191tie
Following instructionsGemini 3.1 Pro
- LiveBench · Instruction FollowingGemini 3.1 Pro by 11.979.167.2+11.9
- IFBenchGemini 3.1 Pro by 1.277.175.9+1.2
Other results16 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- AA-OmniscienceGemini 3.1 Pro by 61.431.9-29.5+61.4
- ARC-AGI-1Gemini 3.1 Pro by 46.59851.5+46.5
- livebench_languageGemini 3.1 Pro by 22.985.462.5+22.9
- Terminal-Bench 2.0Gemini 3.1 Pro by 22.268.546.3+22.2
- τ²-Bench Telecom (AA run)Gemini 3.1 Pro by 19.695.676+19.6
- ToolathlonGemini 3.1 Pro by 13.348.835.5+13.3
- Artificial Analysis Coding IndexGemini 3.1 Pro by 12.768.856.1+12.7
- livebench_data_analysisGemini 3.1 Pro by 10.978.567.6+10.9
- frontiermath_tier_4_v1Gemini 3.1 Pro by 10.416.76.3+10.4
- LiveBenchGemini 3.1 Pro by 9.879.970.1+9.8
- AA IntelligenceGemini 3.1 Pro by 929.720.7+9
- AA Agentic IndexGPT-5.4 nano by 7.410.317.7+7.4
- vectara_factual_consistencyGPT-5.4 nano by 7.389.696.9+7.3
- Finance Agent v2Gemini 3.1 Pro by 4.84338.2+4.8
- vectara_avg_summary_lengthGPT-5.4 nano by 36.7 rating points107.7144.4+36.7 rating
- vectara_answer_ratetie99.4100tie
Questions people ask
Which is better, Gemini 3.1 Pro or GPT-5.4 nano?
Gemini 3.1 Pro wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. GPT-5.4 nano wins none, but costs 90% less.
Which is better for coding?
Gemini 3.1 Pro. It wins 5 of the 6 coding tests both models report; GPT-5.4 nano wins 1.
Which is cheaper?
Gemini 3.1 Pro costs $2.00 per million input tokens and $12.00 per million output tokens; GPT-5.4 nano costs $0.20 and $1.25. That makes GPT-5.4 nano about 90% cheaper for the same work.
How do you compare the two?
We use the 46 benchmark tests both models have published scores on. The verdict counts the 30 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 16 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.