Gemini 3 Flash Preview vs GPT-5.1 Codex mini
Wins 6 of 7 areas
Coding · Agents · Reasoning · Images and charts · Long documents · Following instructions
Wins 0 of 7 areas
—
Gemini 3 Flash Preview is the stronger all-rounder.GPT-5.1 Codex mini is cheaper.
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.
- CodingWriting and fixing software30Gemini 3 Flash Preview3 of 3 tests
- ReasoningHard problems that need careful thinking30Gemini 3 Flash Preview3 of 3 tests
- AgentsCarrying out multi-step tasks on its own10Gemini 3 Flash Preview1 of 1 test
- Images and chartsUnderstanding pictures, charts and video10Gemini 3 Flash Preview1 of 1 test
- Long documentsFinding answers in very long texts10Gemini 3 Flash Preview1 of 1 test
- Following instructionsDoing exactly what it is asked10Gemini 3 Flash Preview1 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 36% 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 Gemini 3 Flash Preview pulls ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+30.2points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+18.1points ahead
- Reasons across sets of long documentsAA-LCR+11.3points ahead
Where GPT-5.1 Codex mini pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+41.4points 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.
CodingGemini 3 Flash Preview
- LMArena · WebDevGemini 3 Flash Preview by 196 rating points14391243+196 rating
- SciCodeGemini 3 Flash Preview by 850.642.6+8
- Terminal-Bench HardGemini 3 Flash Preview by 5.338.633.3+5.3
ReasoningGemini 3 Flash Preview
- Humanity's Last ExamGemini 3 Flash Preview by 18.136.618.5+18.1
- CritPtGemini 3 Flash Preview by 8.68.60+8.6
- GPQA DiamondGemini 3 Flash Preview by 8.589.881.3+8.5
FactsEven
- AA-Omniscience · Non-hallucinationGPT-5.1 Codex mini by 41.4748.4+41.4
- AA-Omniscience · AccuracyGemini 3 Flash Preview by 30.253.423.2+30.2
Images and chartsGemini 3 Flash Preview
- MMMU-ProGemini 3 Flash Preview by 10.979.969+10.9
Long documentsGemini 3 Flash Preview
- AA-LCRGemini 3 Flash Preview by 11.37866.7+11.3
Following instructionsGemini 3 Flash Preview
- IFBenchGemini 3 Flash Preview by 10.17867.9+10.1
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 3 Flash Preview by 57.699.742.1+57.6
- AA-OmniscienceGemini 3 Flash Preview by 26.510.1-16.4+26.5
- τ²-Bench Telecom (AA run)Gemini 3 Flash Preview by 17.580.462.9+17.5
- AA Agentic IndexGemini 3 Flash Preview by 1149.738.7+11
- Artificial Analysis Coding IndexGemini 3 Flash Preview by 6.242.636.4+6.2
- AA IntelligenceGemini 3 Flash Preview by 5.926.320.4+5.9
Questions people ask
Which is better, Gemini 3 Flash Preview or GPT-5.1 Codex mini?
Gemini 3 Flash Preview wins six of the seven areas where both have results: coding, agents, reasoning, images and charts, long documents and following instructions. GPT-5.1 Codex mini wins none, but costs 36% less. They are level on facts.
Which is better for coding?
Gemini 3 Flash Preview. It wins 3 of the 3 coding tests both models report; GPT-5.1 Codex mini wins none.
Which is cheaper?
Gemini 3 Flash Preview costs $0.50 per million input tokens and $3.00 per million output tokens; GPT-5.1 Codex mini costs $0.25 and $2.00. That makes GPT-5.1 Codex mini about 36% 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.