Gemini 3.1 Pro vs GPT-5.2
Wins 6 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Following instructions
Wins 2 of 8 areas
Math · Long documents
Gemini 3.1 Pro is the stronger all-rounder.GPT-5.2 is better at math.
Scores updated · 68 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 6 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video50Gemini 3.1 Pro5 of 6 tests · 1 tie
- CodingWriting and fixing software51Gemini 3.1 Pro5 of 8 tests · 2 ties
- AgentsCarrying out multi-step tasks on its own41Gemini 3.1 Pro4 of 5 tests
- FactsGetting facts right instead of making them up30Gemini 3.1 Pro3 of 4 tests · 1 tie
- Following instructionsDoing exactly what it is asked20Gemini 3.1 Pro2 of 2 tests
- MathCompetition and research-level math05GPT-5.25 of 6 tests · 1 tie
- Long documentsFinding answers in very long texts01GPT-5.21 of 2 tests · 1 tie
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.
Gemini 3.1 Pro costs 11% 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
- Short factual questions, answered correctlySimpleQA Verified+36.4points ahead
- Common-sense trick questionsSimpleBench+33.8points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+30.3points ahead
Where GPT-5.2 pulls ahead
- Finds one of eight look-alike replies in a long chatMRCR v2 8 needle 128k (average)+57.5points ahead
- Real work tasks from 44 professionsGDPVal+33.6points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+7.7points ahead
Every test, side by side
All 68 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGemini 3.1 Pro
- SWE-bench ProGemini 3.1 Pro by 24.354.229.9+24.3
- SWE-bench MultilingualGemini 3.1 Pro by 10.276.966.7+10.2
- Terminal-Bench HardGemini 3.1 Pro by 6.853.847+6.8
- SciCodeGemini 3.1 Pro by 6.658.752.1+6.6
- LiveBench · Agentic CodingGPT-5.2 by 6.244.150.3+6.2
- LMArena · WebDevGemini 3.1 Pro by 31 rating points14471416+31 rating
- SWE-bench Verifiedtie80.680tie
- LiveBench · Codingtie76.576.1tie
AgentsGemini 3.1 Pro
- GDPValGPT-5.2 by 33.614.748.3+33.6
- BrowseCompGemini 3.1 Pro by 20.185.965.8+20.1
- τ-Bench V3 · BankingGemini 3.1 Pro by 10.321.411.1+10.3
- AA ApexAgentsGemini 3.1 Pro by 93223+9
- MCP AtlasGemini 3.1 Pro by 8.669.260.6+8.6
ReasoningGemini 3.1 Pro
- SimpleBenchGemini 3.1 Pro by 33.879.645.8+33.8
- ARC-AGI-2Gemini 3.1 Pro by 24.277.152.9+24.2
- Humanity's Last ExamGemini 3.1 Pro by 9.34737.7+9.3
- CritPtGemini 3.1 Pro by 6.117.711.6+6.1
- GPQA DiamondGemini 3.1 Pro by 3.894.190.3+3.8
- LiveBench · Reasoningtie8483.2tie
FactsGemini 3.1 Pro
- SimpleQA VerifiedGemini 3.1 Pro by 36.473.537.1+36.4
- AA-Omniscience · Non-hallucinationGemini 3.1 Pro by 30.349.118.8+30.3
- AA-Omniscience · AccuracyGemini 3.1 Pro by 10.654.944.3+10.6
- Vectara HHEM hallucination ratelower is bettertie10.410.8tie
Images and chartsGemini 3.1 Pro
- OCRBenchv2Gemini 3.1 Pro by 12.362.850.5+12.3
- MMMU-ProGemini 3.1 Pro by 7.882.474.6+7.8
- MathVistaGemini 3.1 Pro by 7.490.282.8+7.4
- LMArena · VisionGemini 3.1 Pro by 50 rating points12961246+50 rating
- CharXiv (RQ)Gemini 3.1 Pro by 1.283.382.1+1.2
- Video-MMEtie86.786tie
MathGPT-5.2
- FrontierMath Tiers 1-3 (v2)GPT-5.2 by 7.759.667.4+7.7
- IMOAnswerBenchGPT-5.2 by 5.38186.3+5.3
- FrontierMath Tier 4GPT-5.2 by 4.926.831.7+4.9
- HMMT Feb. 2026GPT-5.2 by 2.394.797+2.3
- LiveBench · MathematicsGPT-5.2 by 2.29193.2+2.2
- AIME 2026tie98.398.3tie
Long documentsGPT-5.2
- MRCR v2 8 needle 128k (average)GPT-5.2 by 57.526.383.8+57.5
- AA-LCRtie8282.7tie
Following instructionsGemini 3.1 Pro
- LiveBench · Instruction FollowingGemini 3.1 Pro by 17.379.161.8+17.3
- IFBenchGemini 3.1 Pro by 1.777.175.4+1.7
Other results29 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 Agentic IndexGPT-5.2 by 49.910.360.2+49.9
- LiveCodeBench Pro (Elo)Gemini 3.1 Pro by 494 rating points28872393+494 rating
- AA-OmniscienceGemini 3.1 Pro by 32.831.9-0.9+32.8
- Artificial Analysis Coding IndexGemini 3.1 Pro by 20.168.848.7+20.1
- WorldVQAGemini 3.1 Pro by 16.344.328+16.3
- GDPval-AA (Elo)GPT-5.2 by 145 rating points13171462+145 rating
- Terminal-Bench 2.0Gemini 3.1 Pro by 14.568.554+14.5
- SimpleVQAGemini 3.1 Pro by 14.169.955.8+14.1
- ARC-AGI-1Gemini 3.1 Pro by 11.89886.2+11.8
- τ²-Bench Telecom (AA run)Gemini 3.1 Pro by 10.895.684.8+10.8
- DeepSearchQA (F1)Gemini 3.1 Pro by 10.681.971.3+10.6
- vectara_avg_summary_lengthGPT-5.2 by 78.6 rating points107.7186.3+78.6 rating
- MATH-VisionGemini 3.1 Pro by 6.889.883+6.8
- HLE (with tools)Gemini 3.1 Pro by 5.951.445.5+5.9
- livebench_languageGemini 3.1 Pro by 5.685.479.8+5.6
- LiveBenchGemini 3.1 Pro by 5.179.974.8+5.1
- MotionBenchGemini 3.1 Pro by 5.169.964.8+5.1
- MMLU-ProGemini 3.1 Pro by 4.39186.7+4.3
- MMMLUGemini 3.1 Pro by 392.689.6+3
- matharena_visual_math_overallGemini 3.1 Pro by 2.989.486.5+2.9
- LiveCodeBenchGemini 3.1 Pro by 2.791.789+2.7
- ToolathlonGemini 3.1 Pro by 2.548.846.3+2.5
- frontiermath_tier_4_v1GPT-5.2 by 2.116.718.8+2.1
- HMMT Nov. 2025GPT-5.2 by 194.895.8+1
- AA Intelligencetie29.730.4tie
- vectara_answer_ratetie99.4100tie
- vectara_factual_consistencytie89.689.2tie
- livebench_data_analysistie78.578.2tie
- LongVideoBenchtie76.576.5tie
Questions people ask
Which is better, Gemini 3.1 Pro or GPT-5.2?
Gemini 3.1 Pro wins six of the eight areas we test: coding, agents, reasoning, facts, images and charts and following instructions. GPT-5.2 wins math and long documents.
Which is better for coding?
Gemini 3.1 Pro. It wins 5 of the 8 coding tests both models report; GPT-5.2 wins 1, and 2 are ties.
Which is cheaper?
Gemini 3.1 Pro costs $2.00 per million input tokens and $12.00 per million output tokens; GPT-5.2 costs $1.75 and $14.00. That makes Gemini 3.1 Pro about 11% cheaper for the same work.
How do you compare the two?
We use the 68 benchmark tests both models have published scores on. The verdict counts the 39 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 29 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.