GPT-5 mini vs Grok 4
Wins 3 of 7 areas
Images and charts · Long documents · Following instructions
Wins 4 of 7 areas
Coding · Agents · Reasoning · Facts
Grok 4 wins more areas, narrowly.GPT-5 mini is cheaper and better at images and charts.
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.
- ReasoningHard problems that need careful thinking04Grok 44 of 4 tests
- CodingWriting and fixing software02Grok 42 of 2 tests
- FactsGetting facts right instead of making them up12Grok 42 of 3 tests
- AgentsCarrying out multi-step tasks on its own01Grok 41 of 1 test
- Images and chartsUnderstanding pictures, charts and video10GPT-5 mini1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10GPT-5 mini1 of 1 test
- Following instructionsDoing exactly what it is asked10GPT-5 mini1 of 1 test
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 mini costs 88% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Grok 4 pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+26.3points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+15.5points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+11.5points ahead
Where GPT-5 mini pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+21.7points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+8.1points ahead
- Reasons across sets of long documentsAA-LCR+4.3points ahead
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.
CodingGrok 4
- SciCodeGrok 4 by 6.73945.7+6.7
- Terminal-Bench HardGrok 4 by 4.633.337.9+4.6
ReasoningGrok 4
- ARC-AGI-2Grok 4 by 11.54.415.9+11.5
- Humanity's Last ExamGrok 4 by 5.221.526.7+5.2
- GPQA DiamondGrok 4 by 4.982.887.7+4.9
- CritPtGrok 4 by 202+2
FactsGrok 4
- SimpleQA VerifiedGrok 4 by 26.321.647.9+26.3
- AA-Omniscience · AccuracyGrok 4 by 15.52540.5+15.5
- AA-Omniscience · Non-hallucinationGPT-5 mini by 8.143.635.5+8.1
Images and chartsGPT-5 mini
- MMMU-ProGPT-5 mini by 1.370.168.8+1.3
- LMArena · Visiontie12021210tie
Following instructionsGPT-5 mini
- IFBenchGPT-5 mini by 21.775.453.7+21.7
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.
- HarmBenchGPT-5 mini by 57.497.139.7+57.4
- AIR-Bench 2024GPT-5 mini by 41.385.744.4+41.3
- AA Agentic IndexGrok 4 by 32.68.941.5+32.6
- Artificial Analysis Coding IndexGrok 4 by 24.915.640.5+24.9
- AA-OmniscienceGrok 4 by 19.4-17.32.1+19.4
- ARC-AGI-1Grok 4 by 12.454.366.7+12.4
- matharena_visual_math_overallGPT-5 mini by 8.278.270+8.2
- simple_safety_testsGPT-5 mini by 7.810092.2+7.8
- τ²-Bench Telecom (AA run)Grok 4 by 6.568.474.9+6.5
- AA IntelligenceGrok 4 by 5.716.822.5+5.7
- frontiermath_tier_4_v1GPT-5 mini by 4.26.32.1+4.2
- HMMT Nov. 2025Grok 4 by 4.184.288.3+4.1
- anthropic_red_teamGPT-5 mini by 3.499.195.7+3.4
- bbqGPT-5 mini by 2.696.393.7+2.6
- HMMT 2025Grok 4 by 2.287.890+2.2
- XSTestGPT-5 mini by 1.197.796.6+1.1
- AIME 2025tie91.191.7tie
Questions people ask
Which is better, GPT-5 mini or Grok 4?
Grok 4 wins four of the seven areas where both have results: coding, agents, reasoning and facts. GPT-5 mini wins images and charts, long documents and following instructions, and costs 88% less.
Which is better for coding?
Grok 4. It wins 2 of the 2 coding tests both models report; GPT-5 mini wins none.
Which is cheaper?
GPT-5 mini costs $0.25 per million input tokens and $2.00 per million output tokens; Grok 4 costs $3.00 and $15.00. That makes GPT-5 mini about 88% 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 14 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.