Gemini 3 Flash Preview vs GPT-5.2
Wins 2 of 8 areas
Images and charts · Following instructions
Wins 4 of 8 areas
Coding · Reasoning · Math · Long documents
GPT-5.2 wins more areas, narrowly.Gemini 3 Flash Preview is cheaper and better at images and charts.
Scores updated · 50 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.
- MathCompetition and research-level math04GPT-5.24 of 4 tests
- ReasoningHard problems that need careful thinking13GPT-5.23 of 5 tests · 1 tie
- Long documentsFinding answers in very long texts02GPT-5.22 of 2 tests
- CodingWriting and fixing software23GPT-5.23 of 5 tests
- Images and chartsUnderstanding pictures, charts and video21Gemini 3 Flash Preview2 of 3 tests
- Following instructionsDoing exactly what it is asked10Gemini 3 Flash Preview1 of 1 test
- AgentsCarrying out multi-step tasks on its own22Even2 each
- FactsGetting facts right instead of making them up22Even2 each
Following instructions 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.
Gemini 3 Flash Preview costs 78% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5.2 pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+19.3points ahead
- Finds one of eight look-alike replies in a long chatMRCR v2 8 needle 128k (average)+16.6points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+16.2points ahead
Where Gemini 3 Flash Preview pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+29.7points ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+19.7points ahead
- Common-sense trick questionsSimpleBench+15.3points ahead
Every test, side by side
All 50 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.2
- SWE-bench ProGemini 3 Flash Preview by 19.749.629.9+19.7
- Terminal-Bench HardGPT-5.2 by 8.438.647+8.4
- LMArena · WebDevGemini 3 Flash Preview by 24 rating points14391415+24 rating
- SWE-bench VerifiedGPT-5.2 by 27880+2
- SciCodeGPT-5.2 by 1.550.652.1+1.5
AgentsEven
- GDPValGPT-5.2 by 13.135.248.3+13.1
- τ-Bench V3 · BankingGemini 3 Flash Preview by 9.720.811.1+9.7
- AA ApexAgentsGemini 3 Flash Preview by 4.727.723+4.7
- MCP AtlasGPT-5.2 by 3.257.460.6+3.2
ReasoningGPT-5.2
- ARC-AGI-2GPT-5.2 by 19.333.652.9+19.3
- SimpleBenchGemini 3 Flash Preview by 15.361.145.8+15.3
- CritPtGPT-5.2 by 38.611.6+3
- Humanity's Last ExamGPT-5.2 by 1.136.637.7+1.1
- GPQA Diamondtie89.890.3tie
FactsEven
- SimpleQA VerifiedGemini 3 Flash Preview by 29.766.837.1+29.7
- AA-Omniscience · Non-hallucinationGPT-5.2 by 11.8718.8+11.8
- AA-Omniscience · AccuracyGemini 3 Flash Preview by 9.153.444.3+9.1
- Vectara HHEM hallucination ratelower is betterGPT-5.2 by 2.713.510.8+2.7
Images and chartsGemini 3 Flash Preview
- MMMU-ProGemini 3 Flash Preview by 5.379.974.6+5.3
- LMArena · VisionGemini 3 Flash Preview by 39 rating points12851246+39 rating
- CharXiv (RQ)GPT-5.2 by 1.880.382.1+1.8
MathGPT-5.2
- FrontierMath Tiers 1-3 (v2)GPT-5.2 by 16.251.267.4+16.2
- FrontierMath Tier 4GPT-5.2 by 14.617.131.7+14.6
- HMMT Feb. 2026GPT-5.2 by 7.689.497+7.6
- AIME 2026GPT-5.2 by 1.696.798.3+1.6
Long documentsGPT-5.2
- MRCR v2 8 needle 128k (average)GPT-5.2 by 16.667.283.8+16.6
- AA-LCRGPT-5.2 by 4.77882.7+4.7
Following instructionsGemini 3 Flash Preview
- IFBenchGemini 3 Flash Preview by 2.67875.4+2.6
Other results22 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.
- GDPval-AA (Elo)GPT-5.2 by 258 rating points12041462+258 rating
- ScreenSpot-Pro (No tools)GPT-5.2 by 17.269.186.3+17.2
- frontiermath_tier_4_v1GPT-5.2 by 14.64.218.8+14.6
- AA-OmniscienceGemini 3 Flash Preview by 1110.1-0.9+11
- AA Agentic IndexGPT-5.2 by 10.549.760.2+10.5
- vectara_avg_summary_lengthGPT-5.2 by 96.1 rating points90.2186.3+96.1 rating
- swe_bench_bashGemini 3 Flash Preview by 6.875.869+6.8
- Terminal-Bench 2.0GPT-5.2 by 6.447.654+6.4
- Artificial Analysis Coding IndexGPT-5.2 by 6.142.648.7+6.1
- τ²-Bench Telecom (AA run)GPT-5.2 by 4.480.484.8+4.4
- AA IntelligenceGPT-5.2 by 4.126.330.4+4.1
- ToolathlonGemini 3 Flash Preview by 3.149.446.3+3.1
- vectara_factual_consistencyGPT-5.2 by 2.786.589.2+2.7
- HMMT Feb. 2025GPT-5.2 by 2.597.5100+2.5
- HMMT Nov. 2025GPT-5.2 by 2.593.395.8+2.5
- LiveBenchGPT-5.2 by 2.472.474.8+2.4
- MMMLUGemini 3 Flash Preview by 2.291.889.6+2.2
- ARC-AGI-1GPT-5.2 by 1.584.786.2+1.5
- VideoMMMUGemini 3 Flash Preview by 186.985.9+1
- matharena_visual_math_overalltie85.886.5tie
- AIME 2025tie99.7100tie
- vectara_answer_ratetie99.8100tie
Questions people ask
Which is better, Gemini 3 Flash Preview or GPT-5.2?
GPT-5.2 wins four of the eight areas we test: coding, reasoning, math and long documents. Gemini 3 Flash Preview wins images and charts and following instructions, and costs 78% less. They are level on agents and facts.
Which is better for coding?
GPT-5.2. It wins 3 of the 5 coding tests both models report; Gemini 3 Flash Preview wins 2.
Which is cheaper?
Gemini 3 Flash Preview costs $0.50 per million input tokens and $3.00 per million output tokens; GPT-5.2 costs $1.75 and $14.00. That makes Gemini 3 Flash Preview about 78% cheaper for the same work.
How do you compare the two?
We use the 50 benchmark tests both models have published scores on. The verdict counts the 28 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 22 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.