GPT-4o vs o3
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Images and charts · Long documents · Following instructions
o3 is the stronger all-rounder.
Scores updated · 49 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. 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 thinking04o34 of 4 tests
- Images and chartsUnderstanding pictures, charts and video04o34 of 4 tests
- CodingWriting and fixing software03o33 of 3 tests
- Long documentsFinding answers in very long texts02o32 of 2 tests
- Following instructionsDoing exactly what it is asked02o32 of 2 tests
- FactsGetting facts right instead of making them up12o32 of 3 tests
- AgentsCarrying out multi-step tasks on its own01o31 of 1 test
Agents 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 50% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where o3 pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+35.9points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+35.4points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+30.1points ahead
Where GPT-4o pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+50.2points ahead
Every test, side by side
All 49 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Codingo3
- SWE-bench Verifiedo3 by 35.933.269.1+35.9
- Terminal-Bench Hardo3 by 28.88.337.1+28.8
- SciCodeo3 by 7.733.341+7.7
Reasoningo3
- GPQA Diamondo3 by 30.152.682.7+30.1
- Humanity's Last Examo3 by 18.31.820.1+18.3
- ARC-AGI-2o3 by 6.506.5+6.5
- CritPto3 by 1.101.1+1.1
Factso3
- AA-Omniscience · Non-hallucinationGPT-4o by 50.262.111.9+50.2
- SimpleQA Verifiedo3 by 23.42649.4+23.4
- AA-Omniscience · Accuracyo3 by 14.923.738.6+14.9
Images and chartso3
- MathVistao3 by 25.461.486.8+25.4
- CharXiv (RQ)o3 by 19.858.878.6+19.8
- MMMU-Proo3 by 13.856.370.1+13.8
- MMMUo3 by 10.772.282.9+10.7
Long documentso3
- AA-LCRo3 by 18.75674.7+18.7
- LongBench v2o3 by 10.748.158.8+10.7
Following instructionso3
- IFBencho3 by 35.43671.4+35.4
- Multi-Challengeo3 by 20.140.360.4+20.1
Other results30 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.
- AIME 2024o3 by 82.39.391.6+82.3
- AIME 2025o3 by 74.711.786.4+74.7
- livecodebench_hardo3 by 61.54.566+61.5
- livecodebench_mediumo3 by 57.732.189.8+57.7
- ARC-AGI-1o3 by 56.34.560.8+56.3
- τ²-Bench Telecom (AA run)o3 by 51.828.980.7+51.8
- Aider-Polygloto3 by 50.630.781.3+50.6
- LiveCodeBencho3 by 46.438.384.7+46.4
- metr_hcasto3 by 39.72564.7+39.7
- ERQAo3 by 28.835.264+28.8
- AA Agentic Indexo3 by 27.78.436.1+27.7
- MATH-500 (EM)o3 by 23.574.698.1+23.5
- AIR-Bench 2024o3 by 22.162.484.5+22.1
- VideoMMMUo3 by 22.161.283.3+22.1
- Tau2 airlineo3 by 19.345.564.8+19.3
- τ²-Bench (Retail)o3 by 16.863.480.2+16.8
- livecodebench_easyo3 by 16.682.599.1+16.6
- HarmBencho3 by 15.582.998.4+15.5
- metr_re_bencho3 by 15015+15
- Artificial Analysis Coding Indexo3 by 14.224.238.4+14.2
- TAU-bench (retail)o3 by 13.660.373.9+13.6
- AA Intelligenceo3 by 12.97.320.2+12.9
- SimpleQAo3 by 11.238.249.4+11.2
- MMLU-Proo3 by 10.374.785+10.3
- TAU-bench (airline)o3 by 9.242.852+9.2
- AA-OmniscienceGPT-4o by 5-10.5-15.6+5
- bbqo3 by 2.895.197.9+2.8
- anthropic_red_teamtie99.198.3tie
- simple_safety_teststie98.599tie
- XSTesttie97.397.3tie
Questions people ask
Which is better, GPT-4o or o3?
o3 wins all seven areas where both have results: coding, agents, reasoning, facts, images and charts, long documents and following instructions. GPT-4o wins none.
Which is better for coding?
o3. It wins 3 of the 3 coding tests both models report; GPT-4o wins none.
Which is cheaper?
GPT-4o costs $5.00 per million input tokens and $15.00 per million output tokens; o3 costs $2.00 and $8.00. That makes o3 about 50% cheaper for the same work.
How do you compare the two?
We use the 49 benchmark tests both models have published scores on. The verdict counts the 19 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 30 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.