GPT-5 mini vs GPT-5 nano
Wins 7 of 8 areas
Coding · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
GPT-5 mini is the stronger all-rounder.GPT-5 nano is cheaper.
Scores updated · 39 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 thinking30GPT-5 mini3 of 4 tests · 1 tie
- CodingWriting and fixing software30GPT-5 mini3 of 3 tests
- FactsGetting facts right instead of making them up31GPT-5 mini3 of 4 tests
- Images and chartsUnderstanding pictures, charts and video20GPT-5 mini2 of 2 tests
- MathCompetition and research-level math20GPT-5 mini2 of 2 tests
- Long documentsFinding answers in very long texts10GPT-5 mini1 of 1 test
- Following instructionsDoing exactly what it is asked10GPT-5 mini1 of 1 test
- AgentsCarrying out multi-step tasks on its own00Even0 each · 1 tie
Agents, 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.
GPT-5 nano costs 80% 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 GPT-5 mini pulls ahead
- Reasons across sets of long documentsAA-LCR+27.3points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+26.7points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+25points ahead
Where GPT-5 nano pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 39 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5 mini
- SWE-bench VerifiedGPT-5 mini by 2559.834.8+25
- Terminal-Bench HardGPT-5 mini by 21.233.312.1+21.2
- SciCodeGPT-5 mini by 2.43936.6+2.4
ReasoningGPT-5 mini
- GPQA DiamondGPT-5 mini by 15.282.867.6+15.2
- Humanity's Last ExamGPT-5 mini by 1221.59.5+12
- ARC-AGI-2GPT-5 mini by 1.84.42.6+1.8
- CritPttie00tie
FactsGPT-5 mini
- SimpleQA VerifiedGPT-5 mini by 9.921.611.7+9.9
- AA-Omniscience · AccuracyGPT-5 mini by 5.92519.1+5.9
- AA-Omniscience · Non-hallucinationGPT-5 mini by 2.643.641+2.6
- Vectara HHEM hallucination ratelower is betterGPT-5 nano by 2.412.910.5+2.4
Images and chartsGPT-5 mini
- MMMU-ProGPT-5 mini by 9.170.161+9.1
- LMArena · VisionGPT-5 mini by 43 rating points12021159+43 rating
MathGPT-5 mini
- FrontierMath Tiers 1-3 (v2)GPT-5 mini by 26.746.720+26.7
- FrontierMath Tier 4GPT-5 mini by 9.812.22.4+9.8
Following instructionsGPT-5 mini
- IFBenchGPT-5 mini by 7.875.467.6+7.8
Other results21 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.
- ARC-AGI-1GPT-5 mini by 37.654.316.7+37.6
- τ²-Bench Telecom (AA run)GPT-5 mini by 31.968.436.5+31.9
- swe_bench_bashGPT-5 mini by 2559.834.8+25
- AA Agentic IndexGPT-5 nano by 14.38.923.2+14.3
- FrontierMath (overall)GPT-5 mini by 12.522.19.6+12.5
- HMMT 2025GPT-5 mini by 12.287.875.6+12.2
- AA-OmniscienceGPT-5 mini by 11.4-17.3-28.7+11.4
- vectara_avg_summary_lengthGPT-5 mini by 64 rating points169.7105.7+64 rating
- AIME 2025GPT-5 mini by 5.991.185.2+5.9
- Artificial Analysis Coding IndexGPT-5 nano by 4.715.620.3+4.7
- frontiermath_tier_4_v1GPT-5 mini by 4.26.32.1+4.2
- AA IntelligenceGPT-5 mini by 3.816.813+3.8
- HMMT Nov. 2025GPT-5 mini by 2.584.281.7+2.5
- vectara_factual_consistencyGPT-5 nano by 2.487.189.5+2.4
- AIR-Bench 2024GPT-5 nano by 2.185.787.8+2.1
- bbqGPT-5 nano by 1.396.397.6+1.3
- HarmBenchGPT-5 nano by 1.397.198.4+1.3
- anthropic_red_teamtie99.199.5tie
- vectara_answer_ratetie99.9100tie
- XSTesttie97.797.6tie
- simple_safety_teststie100100tie
Questions people ask
Which is better, GPT-5 mini or GPT-5 nano?
GPT-5 mini wins seven of the eight areas we test: coding, reasoning, facts, images and charts, math, long documents and following instructions. GPT-5 nano wins none, but costs 80% less. They are level on agents.
Which is better for coding?
GPT-5 mini. It wins 3 of the 3 coding tests both models report; GPT-5 nano wins none.
Which is cheaper?
GPT-5 mini costs $0.25 per million input tokens and $2.00 per million output tokens; GPT-5 nano costs $0.05 and $0.40. That makes GPT-5 nano about 80% cheaper for the same work.
How do you compare the two?
We use the 39 benchmark tests both models have published scores on. The verdict counts the 18 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 21 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.