o1 vs o4-mini
Wins 3 of 8 areas
Facts · Long documents · Following instructions
Wins 5 of 8 areas
Coding · Agents · Reasoning · Images and charts · Math
o4-mini is the stronger all-rounder.o1 is better at facts.
Scores updated · 48 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.
- CodingWriting and fixing software03o4-mini3 of 3 tests
- Images and chartsUnderstanding pictures, charts and video03o4-mini3 of 3 tests
- ReasoningHard problems that need careful thinking02o4-mini2 of 3 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own01o4-mini1 of 1 test
- MathCompetition and research-level math01o4-mini1 of 1 test
- FactsGetting facts right instead of making them up30o13 of 3 tests
- Long documentsFinding answers in very long texts10o11 of 1 test
- Following instructionsDoing exactly what it is asked10o11 of 1 test
Agents, math, long documents and following instructions 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.
o4-mini costs 93% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where o4-mini pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+26.8points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+21.4points ahead
- Real work tasks from 44 professionsGDPVal+13.9points ahead
Where o1 pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+22.3points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+10.9points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+9.7points ahead
Every test, side by side
All 48 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Codingo4-mini
- SWE-bench Verifiedo4-mini by 26.841.368.1+26.8
- SciCodeo4-mini by 10.735.846.5+10.7
- Terminal-Bench Hardo4-mini by 2.312.915.2+2.3
Reasoningo4-mini
- Humanity's Last Examo4-mini by 9.5716.5+9.5
- GPQA Diamondo4-mini by 3.774.778.4+3.7
- CritPttie0.30.6tie
Factso1
- SimpleQA Verifiedo1 by 22.341.118.8+22.3
- AA-Omniscience · Non-hallucinationo1 by 10.930.419.5+10.9
- AA-Omniscience · Accuracyo1 by 9.734.524.8+9.7
Images and chartso4-mini
- MathVistao4-mini by 12.571.884.3+12.5
- MMMUo4-mini by 477.681.6+4
- LMArena · Visiono4-mini by 26 rating points11681194+26 rating
Other results32 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.
- AA-Omniscienceo1 by 24.7-11.1-35.7+24.7
- Artificial Analysis Coding Indexo1 by 14.139.725.6+14.1
- AIME 2025o4-mini by 1181.792.7+11
- Aider-Polygloto4-mini by 7.261.768.9+7.2
- τ²-Bench Telecom (AA run)o1 by 762.655.6+7
- AA Agentic Indexo4-mini by 531.136.1+5
- MMLU Language (0-shot) - Yorubao1 by 4.675.470.8+4.6
- MMLU Language (0-shot) - Swahilio1 by 4.185.481.3+4.1
- bbqo1 by 3.397.394+3.3
- MMLU Language (0-shot) - Averageo1 by 2.587.785.2+2.5
- MMLU Language (0-shot) - Hindio1 by 2.488.385.9+2.4
- MMLU Language (0-shot) - Chinese (Simplified)o1 by 288.986.9+2
- MMLU Language (0-shot) - Italiano1 by 289.787.7+2
- MMLU Language (0-shot) - Japaneseo1 by 288.986.9+2
- MMLU Language (0-shot) - Frencho1 by 1.989.387.4+1.9
- MMLU Language (0-shot) - Indonesiano1 by 1.788.686.9+1.7
- MMLU Language (0-shot) - Portuguese (Brazil)o1 by 1.789.587.8+1.7
- AA Intelligenceo4-mini by 1.515.216.7+1.5
- AIR-Bench 2024o1 by 1.58078.5+1.5
- MMLU Language (0-shot) - Koreano1 by 1.588.286.7+1.5
- simple_safety_testso4-mini by 199100+1
- StrongREJECTo1 by 19796+1
- TAU-bench (retail)o4-mini by 170.871.8+1
- TAU-bench (airline)tie5049.2tie
- HarmBenchtie96.397tie
- SimpleQA hallucination ratelower is bettertie0.40.8tie
- XSTesttie9797.4tie
- PersonQA hallucination ratelower is bettertie0.20.5tie
- BBQ Accuracy on Ambiguous Questionstie10.8tie
- BBQ P(not stereotyping | ambiguous question, not unknown)tie0.10.3tie
- anthropic_red_teamtie98.398.2tie
- BBQ Accuracy on Unambiguous Questionstie0.90.9tie
Questions people ask
Which is better, o1 or o4-mini?
o4-mini wins five of the eight areas we test: coding, agents, reasoning, images and charts and math. o1 wins facts, long documents and following instructions.
Which is better for coding?
o4-mini. It wins 3 of the 3 coding tests both models report; o1 wins none.
Which is cheaper?
o1 costs $15.00 per million input tokens and $60.00 per million output tokens; o4-mini costs $1.10 and $4.40. That makes o4-mini about 93% cheaper for the same work.
How do you compare the two?
We use the 48 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 32 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.