GPT-5.4 mini vs Grok 4.3
Wins 3 of 8 areas
Coding · Agents · Long documents
Wins 3 of 8 areas
Reasoning · Math · Following instructions
The two are evenly matched.GPT-5.4 mini is better at coding and agents; Grok 4.3 at following instructions and reasoning.
Scores updated · 31 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.
- CodingWriting and fixing software60GPT-5.4 mini6 of 6 tests
- AgentsCarrying out multi-step tasks on its own50GPT-5.4 mini5 of 5 tests
- Long documentsFinding answers in very long texts10GPT-5.4 mini1 of 1 test
- Following instructionsDoing exactly what it is asked02Grok 4.32 of 2 tests
- ReasoningHard problems that need careful thinking12Grok 4.32 of 4 tests · 1 tie
- MathCompetition and research-level math01Grok 4.31 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
- Images and chartsUnderstanding pictures, charts and video11Even1 each
Math and long documents 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.
Grok 4.3 costs 29% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5.4 mini pulls ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+23.2points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+19.5points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+14.4points ahead
Where Grok 4.3 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+64.4points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+9.1points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+8points 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.
CodingGPT-5.4 mini
- LiveBench · Agentic CodingGPT-5.4 mini by 23.241.718.5+23.2
- Terminal-Bench 2.1GPT-5.4 mini by 19.559.239.7+19.5
- Terminal-Bench HardGPT-5.4 mini by 14.452.337.9+14.4
- LMArena · WebDevGPT-5.4 mini by 40 rating points13971357+40 rating
- SciCodeGPT-5.4 mini by 3.852.148.3+3.8
- LiveBench · CodingGPT-5.4 mini by 1.771.669.9+1.7
AgentsGPT-5.4 mini
- τ-Bench V3 · BankingGPT-5.4 mini by 13.225.612.4+13.2
- AA ApexAgentsGPT-5.4 mini by 11.228.217+11.2
- GDPValGPT-5.4 mini by 3.825.822+3.8
- AA IT-Bench SREGPT-5.4 mini by 2.535.232.7+2.5
- Terminal-Bench 4.0GPT-5.4 mini by 220+2
ReasoningGrok 4.3
- Humanity's Last ExamGrok 4.3 by 9.128.137.2+9.1
- GPQA DiamondGrok 4.3 by 2.687.590.1+2.6
- CritPtGPT-5.4 mini by 2108+2
- LiveBench · Reasoningtie71.370.8tie
FactsEven
- AA-Omniscience · Non-hallucinationGrok 4.3 by 64.49.874.2+64.4
- AA-Omniscience · AccuracyGPT-5.4 mini by 2.737.534.8+2.7
Images and chartsEven
- MMMU-ProGrok 4.3 by 4.873.378.1+4.8
- LMArena · VisionGPT-5.4 mini by 16 rating points12451229+16 rating
Following instructionsGrok 4.3
- IFBenchGrok 4.3 by 873.381.3+8
- LiveBench · Instruction FollowingGrok 4.3 by 359.862.8+3
Other results8 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.
- AA-OmniscienceGrok 4.3 by 36.9-18.918+36.9
- livebench_data_analysisGPT-5.4 mini by 1570.855.8+15
- τ²-Bench Telecom (AA run)Grok 4.3 by 14.483.397.7+14.4
- Artificial Analysis Coding IndexGPT-5.4 mini by 13.956.142.2+13.9
- Finance Agent v2GPT-5.4 mini by 7.745.437.7+7.7
- livebench_languageGrok 4.3 by 2.67173.6+2.6
- AA Agentic IndexGPT-5.4 mini by 2.519.717.2+2.5
- AA Intelligencetie24.124.9tie
Questions people ask
Which is better, GPT-5.4 mini or Grok 4.3?
GPT-5.4 mini and Grok 4.3 each win three of the eight areas we test. GPT-5.4 mini wins coding, agents and long documents; Grok 4.3 wins reasoning, math and following instructions. Grok 4.3 costs 29% less. They are level on facts and images and charts.
Which is better for coding?
GPT-5.4 mini. It wins 6 of the 6 coding tests both models report; Grok 4.3 wins none.
Which is cheaper?
GPT-5.4 mini costs $0.75 per million input tokens and $4.50 per million output tokens; Grok 4.3 costs $1.25 and $2.50. That makes Grok 4.3 about 29% 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 23 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 8 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.