GPT-5 vs o1
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
GPT-5 is the stronger all-rounder.
Scores updated · 33 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 thinking40GPT-54 of 4 tests
- CodingWriting and fixing software30GPT-53 of 3 tests
- Images and chartsUnderstanding pictures, charts and video20GPT-52 of 2 tests
- FactsGetting facts right instead of making them up21GPT-52 of 3 tests
- AgentsCarrying out multi-step tasks on its own10GPT-51 of 1 test
- MathCompetition and research-level math10GPT-51 of 1 test
- Long documentsFinding answers in very long texts10GPT-51 of 1 test
- Following instructionsDoing exactly what it is asked10GPT-51 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.
GPT-5 costs 85% 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 pulls ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+40.7points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+33.6points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+21.5points ahead
Where o1 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+12.6points ahead
Every test, side by side
All 33 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5
- SWE-bench VerifiedGPT-5 by 33.674.941.3+33.6
- Terminal-Bench HardGPT-5 by 19.732.612.9+19.7
- SciCodeGPT-5 by 7.142.935.8+7.1
ReasoningGPT-5
- Humanity's Last ExamGPT-5 by 21.528.57+21.5
- SimpleBenchGPT-5 by 16.656.740.1+16.6
- GPQA DiamondGPT-5 by 10.785.474.7+10.7
- CritPtGPT-5 by 5.45.70.3+5.4
FactsGPT-5
- AA-Omniscience · Non-hallucinationo1 by 12.617.830.4+12.6
- SimpleQA VerifiedGPT-5 by 950.141.1+9
- AA-Omniscience · AccuracyGPT-5 by 5.840.334.5+5.8
Images and chartsGPT-5
- MMMUGPT-5 by 6.684.277.6+6.6
- LMArena · VisionGPT-5 by 40 rating points12081168+40 rating
Other results17 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.
- metr_re_benchGPT-5 by 40400+40
- Aider-PolyglotGPT-5 by 26.38861.7+26.3
- metr_hcastGPT-5 by 25.868.442.6+25.8
- τ²-Bench Telecom (AA run)GPT-5 by 22.284.862.6+22.2
- FrontierMath (overall)GPT-5 by 20.826.35.5+20.8
- AIME 2025GPT-5 by 12.994.681.7+12.9
- AA IntelligenceGPT-5 by 7.82315.2+7.8
- AIR-Bench 2024GPT-5 by 7.787.780+7.7
- HumanEvalGPT-5 by 5.393.488.1+5.3
- AA Agentic Indexo1 by 4.626.531.1+4.6
- AA-OmniscienceGPT-5 by 2.3-8.7-11.1+2.3
- Artificial Analysis Coding Indexo1 by 1.937.839.7+1.9
- HarmBenchGPT-5 by 1.397.696.3+1.3
- anthropic_red_teamtie99.198.3tie
- simple_safety_teststie99.899tie
- bbqtie96.897.3tie
- XSTesttie97.197tie
Questions people ask
Which is better, GPT-5 or o1?
GPT-5 wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. o1 wins none.
Which is better for coding?
GPT-5. It wins 3 of the 3 coding tests both models report; o1 wins none.
Which is cheaper?
GPT-5 costs $1.25 per million input tokens and $10.00 per million output tokens; o1 costs $15.00 and $60.00. That makes GPT-5 about 85% cheaper for the same work.
How do you compare the two?
We use the 33 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 17 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.