Gemini 2.5 Pro vs o3
Wins 0 of 8 areas
—
Wins 6 of 8 areas
Coding · Agents · Reasoning · Math · Long documents · Following instructions
o3 is the stronger all-rounder.
Scores updated · 53 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 thinking13o33 of 5 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own02o32 of 2 tests
- Following instructionsDoing exactly what it is asked02o32 of 2 tests
- CodingWriting and fixing software12o32 of 3 tests
- Long documentsFinding answers in very long texts01o31 of 2 tests · 1 tie
- MathCompetition and research-level math01o31 of 1 test
- FactsGetting facts right instead of making them up11Even1 each · 1 tie
- Images and chartsUnderstanding pictures, charts and video22Even2 each
Math 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.
o3 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 o3 pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+39.8points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+22.7points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+10.6points ahead
Where Gemini 2.5 Pro pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+6.6points ahead
- Code for real scientific research problemsSciCode+5.3points ahead
- Harder college exam questions with imagesMMMU-Pro+4.8points ahead
Every test, side by side
All 53 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Codingo3
- Terminal-Bench Hardo3 by 10.626.537.1+10.6
- SWE-bench Verifiedo3 by 5.963.269.1+5.9
- SciCodeGemini 2.5 Pro by 5.346.341+5.3
Reasoningo3
- Humanity's Last Examo3 by 2.11820.1+2.1
- ARC-AGI-2o3 by 1.64.96.5+1.6
- CritPtGemini 2.5 Pro by 1.52.61.1+1.5
- SimpleBencho3 by 1.551.653.1+1.5
- GPQA Diamondtie83.682.7tie
FactsEven
- SimpleQA VerifiedGemini 2.5 Pro by 6.65649.4+6.6
- AA-Omniscience · Non-hallucinationo3 by 2.89.111.9+2.8
- AA-Omniscience · Accuracytie3938.6tie
Images and chartsEven
- LMArena · VisionGemini 2.5 Pro by 49 rating points12631214+49 rating
- MMMU-ProGemini 2.5 Pro by 4.874.970.1+4.8
- MMMUo3 by 3.379.682.9+3.3
- MathVistao3 by 2.983.986.8+2.9
Following instructionso3
- IFBencho3 by 22.748.771.4+22.7
- Multi-Challengeo3 by 6.853.660.4+6.8
Other results31 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.
- HarmBencho3 by 3365.498.4+33
- AA Agentic Indexo3 by 32.63.536.1+32.6
- τ²-Bench Telecom (AA run)o3 by 26.654.180.7+26.6
- ARC-AGI-1o3 by 23.83760.8+23.8
- OpenAI-MRCR (128k)Gemini 2.5 Pro by 20.376.856.5+20.3
- imo_2025Gemini 2.5 Pro by 14.931.616.7+14.9
- AIR-Bench 2024o3 by 10.973.684.5+10.9
- LiveCodeBencho3 by 10.574.284.7+10.5
- Artificial Analysis Coding IndexGemini 2.5 Pro by 8.346.738.4+8.3
- TAU-bench (retail)o3 by 6.96773.9+6.9
- livecodebench_hardo3 by 6.659.466+6.6
- AA Intelligenceo3 by 5.21520.2+5.2
- Aider-Polygloto3 by 4.876.581.3+4.8
- ZebraLogico3 by 4.291.695.8+4.2
- AIME 2025o3 by 3.48386.4+3.4
- HMMT Feb. 2025Gemini 2.5 Pro by 3.380.877.5+3.3
- frontiermath_tier_4_v1Gemini 2.5 Pro by 2.14.22.1+2.1
- simple_safety_testso3 by 29799+2
- TAU-bench (airline)o3 by 25052+2
- bbqo3 by 1.596.497.9+1.5
- SimpleQAGemini 2.5 Pro by 1.450.849.4+1.4
- XSTestGemini 2.5 Pro by 1.498.797.3+1.4
- LiveCodeBench (24/8~25/5)Gemini 2.5 Pro by 1.377.175.8+1.3
- anthropic_red_teamGemini 2.5 Pro by 1.299.598.3+1.2
- MMLU-ProGemini 2.5 Pro by 18685+1
- AA-Omnisciencetie-16.4-15.6tie
- livecodebench_mediumtie90.689.8tie
- MATH-500 (EM)tie98.898.1tie
- AIME 2024tie9291.6tie
- livecodebench_easytie98.899.1tie
- VideoMMMUtie83.683.3tie
Questions people ask
Which is better, Gemini 2.5 Pro or o3?
o3 wins six of the eight areas we test: coding, agents, reasoning, math, long documents and following instructions. Gemini 2.5 Pro wins none. They are level on facts and images and charts.
Which is better for coding?
o3. It wins 2 of the 3 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; o3 costs $2.00 and $8.00. That makes o3 about 11% cheaper for the same work.
How do you compare the two?
We use the 53 benchmark tests both models have published scores on. The verdict counts the 22 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 31 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.