Gemini 2.5 Pro vs GPT-5.1 Codex mini
Wins 3 of 7 areas
Reasoning · Images and charts · Long documents
Wins 3 of 7 areas
Coding · Agents · Following instructions
The two are evenly matched.Gemini 2.5 Pro is better at reasoning and images and charts; GPT-5.1 Codex mini at coding and agents.
Scores updated · 18 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.
- ReasoningHard problems that need careful thinking20Gemini 2.5 Pro2 of 3 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video10Gemini 2.5 Pro1 of 1 test
- Long documentsFinding answers in very long texts10Gemini 2.5 Pro1 of 1 test
- CodingWriting and fixing software12GPT-5.1 Codex mini2 of 3 tests
- AgentsCarrying out multi-step tasks on its own01GPT-5.1 Codex mini1 of 1 test
- Following instructionsDoing exactly what it is asked01GPT-5.1 Codex mini1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Agents, images and charts, 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-5.1 Codex mini costs 80% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Gemini 2.5 Pro pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+15.9points ahead
- Harder college exam questions with imagesMMMU-Pro+5.9points ahead
- Code for real scientific research problemsSciCode+3.7points ahead
Where GPT-5.1 Codex mini pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+39.3points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+19.2points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+6.8points ahead
Every test, side by side
All 18 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.1 Codex mini
- Terminal-Bench HardGPT-5.1 Codex mini by 6.826.533.3+6.8
- SciCodeGemini 2.5 Pro by 3.746.342.6+3.7
- LMArena · WebDevGPT-5.1 Codex mini by 19 rating points12241243+19 rating
ReasoningGemini 2.5 Pro
- CritPtGemini 2.5 Pro by 2.62.60+2.6
- GPQA DiamondGemini 2.5 Pro by 2.383.681.3+2.3
- Humanity's Last Examtie1818.5tie
FactsEven
- AA-Omniscience · Non-hallucinationGPT-5.1 Codex mini by 39.39.148.4+39.3
- AA-Omniscience · AccuracyGemini 2.5 Pro by 15.93923.2+15.9
Images and chartsGemini 2.5 Pro
- MMMU-ProGemini 2.5 Pro by 5.974.969+5.9
Following instructionsGPT-5.1 Codex mini
- IFBenchGPT-5.1 Codex mini by 19.248.767.9+19.2
Other results6 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 2025Gemini 2.5 Pro by 40.98342.1+40.9
- AA Agentic IndexGPT-5.1 Codex mini by 35.23.538.7+35.2
- Artificial Analysis Coding IndexGemini 2.5 Pro by 10.346.736.4+10.3
- τ²-Bench Telecom (AA run)GPT-5.1 Codex mini by 8.854.162.9+8.8
- AA IntelligenceGPT-5.1 Codex mini by 5.41520.4+5.4
- AA-Omnisciencetie-16.4-16.4tie
Questions people ask
Which is better, Gemini 2.5 Pro or GPT-5.1 Codex mini?
Gemini 2.5 Pro and GPT-5.1 Codex mini each win three of the seven areas where both have results. Gemini 2.5 Pro wins reasoning, images and charts and long documents; GPT-5.1 Codex mini wins coding, agents and following instructions. GPT-5.1 Codex mini costs 80% less. They are level on facts.
Which is better for coding?
GPT-5.1 Codex mini. It wins 2 of the 3 coding tests both models report; Gemini 2.5 Pro wins 1.
Which is cheaper?
Gemini 2.5 Pro costs $1.25 per million input tokens and $10.00 per million output tokens; GPT-5.1 Codex mini costs $0.25 and $2.00. That makes GPT-5.1 Codex mini about 80% cheaper for the same work.
How do you compare the two?
We use the 18 benchmark tests both models have published scores on. The verdict counts the 12 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 6 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.