GPT-4o vs Step3 VL 10B
Wins 3 of 7 areas
Coding · Facts · Long documents
Wins 3 of 7 areas
Reasoning · Images and charts · Following instructions
The two are evenly matched.GPT-4o is better at coding and facts; Step3 VL 10B at images and charts and reasoning.
Scores updated · 26 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.
- Images and chartsUnderstanding pictures, charts and video05Step3 VL 10B5 of 5 tests
- ReasoningHard problems that need careful thinking02Step3 VL 10B2 of 3 tests · 1 tie
- Following instructionsDoing exactly what it is asked02Step3 VL 10B2 of 2 tests
- CodingWriting and fixing software20GPT-4o2 of 2 tests
- FactsGetting facts right instead of making them up20GPT-4o2 of 2 tests
- Long documentsFinding answers in very long texts10GPT-4o1 of 1 test
- AgentsCarrying out multi-step tasks on its own00Even0 each · 1 tie
Agents and long documents rest on a single test each.
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 Step3 VL 10B pulls ahead
- Math problems shown in pictures and chartsMathVista+22.6points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+22.3points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+16.4points ahead
Where GPT-4o pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+45.4points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+10.4points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+3points ahead
Every test, side by side
All 26 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-4o
- Terminal-Bench HardGPT-4o by 38.35.3+3
- SciCodeGPT-4o by 2.233.331.1+2.2
ReasoningStep3 VL 10B
- GPQA DiamondStep3 VL 10B by 16.452.669+16.4
- Humanity's Last ExamStep3 VL 10B by 91.810.8+9
- CritPttie00tie
FactsGPT-4o
- AA-Omniscience · Non-hallucinationGPT-4o by 45.462.116.7+45.4
- AA-Omniscience · AccuracyGPT-4o by 10.423.713.3+10.4
Images and chartsStep3 VL 10B
Following instructionsStep3 VL 10B
- Multi-ChallengeStep3 VL 10B by 22.340.362.6+22.3
- IFBenchStep3 VL 10B by 14.23650.2+14.2
Other results10 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 2025Step3 VL 10B by 7611.787.7+76
- ScreenSpot-V2Step3 VL 10B by 74.518.192.6+74.5
- AA-OmniscienceGPT-4o by 48.5-10.5-59+48.5
- MATH-VisionStep3 VL 10B by 40.430.470.8+40.4
- LiveCodeBenchStep3 VL 10B by 38.138.376.4+38.1
- τ²-Bench Telecom (AA run)GPT-4o by 12.828.916.1+12.8
- Artificial Analysis Coding IndexGPT-4o by 10.324.213.9+10.3
- AI2DGPT-4o by 4.894.289.3+4.8
- AA Agentic IndexGPT-4o by 38.45.4+3
- AA Intelligencetie7.37.7tie
Questions people ask
Which is better, GPT-4o or Step3 VL 10B?
GPT-4o and Step3 VL 10B each win three of the seven areas where both have results. GPT-4o wins coding, facts and long documents; Step3 VL 10B wins reasoning, images and charts and following instructions. They are level on agents.
Which is better for coding?
GPT-4o. It wins 2 of the 2 coding tests both models report; Step3 VL 10B wins none.
How do you compare the two?
We use the 26 benchmark tests both models have published scores on. The verdict counts the 16 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 10 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.