GPT-4.1 nano vs o4-mini
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Images and charts · Long documents · Following instructions
o4-mini is the stronger all-rounder.GPT-4.1 nano is cheaper.
Scores updated · 31 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 video04o4-mini4 of 4 tests
- ReasoningHard problems that need careful thinking03o4-mini3 of 4 tests · 1 tie
- FactsGetting facts right instead of making them up03o4-mini3 of 3 tests
- CodingWriting and fixing software02o4-mini2 of 2 tests
- Following instructionsDoing exactly what it is asked02o4-mini2 of 2 tests
- AgentsCarrying out multi-step tasks on its own01o4-mini1 of 1 test
- Long documentsFinding answers in very long texts01o4-mini1 of 1 test
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.
GPT-4.1 nano costs 91% 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 o4-mini pulls ahead
- Reasons across sets of long documentsAA-LCR+40.7points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+36.7points ahead
- Reasoning about charts from research papersCharXiv (RQ)+31.5points ahead
Where GPT-4.1 nano pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 31 tests both models report. The winning score is in its model's colour; marks a score checked independently.
Codingo4-mini
- SciCodeo4-mini by 20.625.946.5+20.6
- Terminal-Bench Hardo4-mini by 11.43.815.2+11.4
Reasoningo4-mini
- GPQA Diamondo4-mini by 27.251.278.4+27.2
- Humanity's Last Examo4-mini by 12.73.816.5+12.7
- ARC-AGI-2o4-mini by 6.106.1+6.1
- CritPttie00.6tie
Factso4-mini
- SimpleQA Verifiedo4-mini by 12.8618.8+12.8
- AA-Omniscience · Accuracyo4-mini by 11.113.724.8+11.1
- AA-Omniscience · Non-hallucinationo4-mini by 2.117.419.5+2.1
Images and chartso4-mini
- CharXiv (RQ)o4-mini by 31.540.572+31.5
- MMMU-Proo4-mini by 29.140.169.2+29.1
- MathVistao4-mini by 28.156.284.3+28.1
- MMMUo4-mini by 26.255.481.6+26.2
Following instructionso4-mini
- IFBencho4-mini by 36.73268.7+36.7
- Multi-Challengeo4-mini by 281543+28
Other results14 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.
- Aider-Polygloto4-mini by 59.19.868.9+59.1
- ARC-AGI-1o4-mini by 58.7058.7+58.7
- TAU-bench (retail)o4-mini by 49.222.671.8+49.2
- τ²-Bench Telecom (AA run)o4-mini by 38.317.355.6+38.3
- TAU-bench (airline)o4-mini by 35.21449.2+35.2
- AA-Omniscienceo4-mini by 21.9-57.6-35.7+21.9
- AIR-Bench 2024o4-mini by 1761.578.5+17
- Artificial Analysis Coding Indexo4-mini by 14.511.125.6+14.5
- HarmBencho4-mini by 10.286.897+10.2
- AA Intelligenceo4-mini by 8.97.816.7+8.9
- bbqo4-mini by 6.587.594+6.5
- anthropic_red_teamGPT-4.1 nano by 1.499.698.2+1.4
- XSTesto4-mini by 1.49697.4+1.4
- simple_safety_testso4-mini by 199100+1
Questions people ask
Which is better, GPT-4.1 nano or o4-mini?
o4-mini wins all seven areas where both have results: coding, agents, reasoning, facts, images and charts, long documents and following instructions. GPT-4.1 nano wins none, but costs 91% less.
Which is better for coding?
o4-mini. It wins 2 of the 2 coding tests both models report; GPT-4.1 nano wins none.
Which is cheaper?
GPT-4.1 nano costs $0.10 per million input tokens and $0.40 per million output tokens; o4-mini costs $1.10 and $4.40. That makes GPT-4.1 nano about 91% cheaper for the same work.
How do you compare the two?
We use the 31 benchmark tests both models have published scores on. The verdict counts the 17 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 14 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.