Gemini 2.5 Pro vs GPT-5
Wins 1 of 8 areas
Images and charts
Wins 6 of 8 areas
Coding · Agents · Reasoning · Math · Long documents · Following instructions
GPT-5 is the stronger all-rounder.Gemini 2.5 Pro is better at images and charts.
Scores updated · 60 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 thinking05GPT-55 of 5 tests
- CodingWriting and fixing software14GPT-54 of 5 tests
- AgentsCarrying out multi-step tasks on its own03GPT-53 of 3 tests
- MathCompetition and research-level math02GPT-52 of 2 tests
- Following instructionsDoing exactly what it is asked02GPT-52 of 2 tests
- Long documentsFinding answers in very long texts01GPT-51 of 1 test
- Images and chartsUnderstanding pictures, charts and video21Gemini 2.5 Pro2 of 4 tests · 1 tie
- FactsGetting facts right instead of making them up22Even2 each
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.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5 pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+45points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+30.8points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+24.4points ahead
Where Gemini 2.5 Pro pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+5.9points ahead
- Reads and reasons about text in imagesOCRBenchv2+3.8points ahead
- Code for real scientific research problemsSciCode+3.4points ahead
Every test, side by side
All 60 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5
- LMArena · WebDevGPT-5 by 194 rating points12241418+194 rating
- SWE-bench VerifiedGPT-5 by 11.763.274.9+11.7
- Terminal-Bench 2.1GPT-5 by 6.728.535.2+6.7
- Terminal-Bench HardGPT-5 by 6.126.532.6+6.1
- SciCodeGemini 2.5 Pro by 3.446.342.9+3.4
AgentsGPT-5
- BrowseCompGPT-5 by 459.954.9+45
- GDPValGPT-5 by 21021+21
- τ-Bench V3 · BankingGPT-5 by 12.49.722.1+12.4
ReasoningGPT-5
- Humanity's Last ExamGPT-5 by 10.51828.5+10.5
- SimpleBenchGPT-5 by 5.151.656.7+5.1
- ARC-AGI-2GPT-5 by 54.99.9+5
- CritPtGPT-5 by 3.12.65.7+3.1
- GPQA DiamondGPT-5 by 1.883.685.4+1.8
FactsEven
- AA-Omniscience · Non-hallucinationGPT-5 by 8.79.117.8+8.7
- Vectara HHEM hallucination ratelower is betterGemini 2.5 Pro by 8.1715.1+8.1
- SimpleQA VerifiedGemini 2.5 Pro by 5.95650.1+5.9
- AA-Omniscience · AccuracyGPT-5 by 1.23940.3+1.2
Images and chartsGemini 2.5 Pro
- LMArena · VisionGemini 2.5 Pro by 55 rating points12631208+55 rating
- MMMUGPT-5 by 4.679.684.2+4.6
- OCRBenchv2Gemini 2.5 Pro by 3.859.355.5+3.8
- MMMU-Protie74.974.2tie
MathGPT-5
- FrontierMath Tiers 1-3 (v2)GPT-5 by 30.824.655.4+30.8
- FrontierMath Tier 4GPT-5 by 22021.9+22
Following instructionsGPT-5
- IFBenchGPT-5 by 24.448.773.1+24.4
- Multi-ChallengeGPT-5 by 1653.669.6+16
Other results34 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.
- HarmBenchGPT-5 by 32.265.497.6+32.2
- BrowseComp-ZHGPT-5 by 30.832.263+30.8
- τ²-Bench Telecom (AA run)GPT-5 by 30.754.184.8+30.7
- ARC-AGI-1GPT-5 by 28.73765.7+28.7
- τ²-BenchGPT-5 by 28.45482.4+28.4
- HMMT 2025GPT-5 by 27.665.793.3+27.6
- AA Agentic IndexGPT-5 by 233.526.5+23
- xbench-DeepSearchGPT-5 by 21.85677.8+21.8
- FinSearchComp-globalGPT-5 by 21.342.663.9+21.3
- Terminal-BenchGPT-5 by 18.525.343.8+18.5
- GAIA (text only)GPT-5 by 16.260.276.4+16.2
- ArtifactsBenchGPT-5 by 15.357.773+15.3
- AIR-Bench 2024GPT-5 by 14.173.687.7+14.1
- AIME 2025GPT-5 by 11.68394.6+11.6
- Aider-PolyglotGPT-5 by 11.576.588+11.5
- LiveCodeBenchGPT-5 by 9.374.283.5+9.3
- HMMT Nov. 2025GPT-5 by 9.28089.2+9.2
- Artificial Analysis Coding IndexGemini 2.5 Pro by 8.946.737.8+8.9
- frontiermath_tier_4_v1GPT-5 by 8.34.212.5+8.3
- vectara_factual_consistencyGemini 2.5 Pro by 8.19384.9+8.1
- AA IntelligenceGPT-5 by 81523+8
- AA-OmniscienceGPT-5 by 7.6-16.4-8.7+7.6
- HMMT Feb. 2025GPT-5 by 7.580.888.3+7.5
- HLE (with tools)GPT-5 by 6.828.435.2+6.8
- imo_2025GPT-5 by 6.531.638.1+6.5
- vectara_avg_summary_lengthGPT-5 by 56.3 rating points106.4162.7+56.3 rating
- simple_safety_testsGPT-5 by 2.89799.8+2.8
- matharena_visual_math_overallGPT-5 by 1.677.278.8+1.6
- XSTestGemini 2.5 Pro by 1.698.797.1+1.6
- MMLU-ProGPT-5 by 1.18687.1+1.1
- VideoMMMUGPT-5 by 183.684.6+1
- vectara_answer_ratetie99.199.9tie
- anthropic_red_teamtie99.599.1tie
- bbqtie96.496.8tie
Questions people ask
Which is better, Gemini 2.5 Pro or GPT-5?
GPT-5 wins six of the eight areas we test: coding, agents, reasoning, math, long documents and following instructions. Gemini 2.5 Pro wins images and charts. They are level on facts.
Which is better for coding?
GPT-5. It wins 4 of the 5 coding tests both models report; Gemini 2.5 Pro wins 1.
Which is cheaper?
Gemini 2.5 Pro costs $1.25 per million input tokens and $10.00 per million output tokens; GPT-5 costs $1.25 and $10.00. For a million tokens read plus a million written, they cost about the same.
How do you compare the two?
We use the 60 benchmark tests both models have published scores on. The verdict counts the 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 34 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.