GPT-5.4 mini vs GPT-5.5
Wins 0 of 8 areas
—
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
GPT-5.5 is the stronger all-rounder.GPT-5.4 mini is cheaper.
Scores updated · 47 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 software07GPT-5.57 of 7 tests
- AgentsCarrying out multi-step tasks on its own07GPT-5.57 of 7 tests
- ReasoningHard problems that need careful thinking05GPT-5.55 of 5 tests
- MathCompetition and research-level math03GPT-5.53 of 3 tests
- FactsGetting facts right instead of making them up13GPT-5.53 of 4 tests
- Images and chartsUnderstanding pictures, charts and video02GPT-5.52 of 2 tests
- Following instructionsDoing exactly what it is asked02GPT-5.52 of 2 tests
- Long documentsFinding answers in very long texts01GPT-5.51 of 1 test
Long documents rests on a single test.
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-5.4 mini 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-5.5 pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+66.1points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+62.7points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+34.1points ahead
Where GPT-5.4 mini pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 47 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.5
- Terminal-Bench 2.1GPT-5.5 by 25.159.284.3+25.1
- LiveBench · Agentic CodingGPT-5.5 by 12.341.754+12.3
- LMArena · WebDevGPT-5.5 by 116 rating points13971513+116 rating
- LiveBench · CodingGPT-5.5 by 10.671.682.2+10.6
- Terminal-Bench HardGPT-5.5 by 8.352.360.6+8.3
- SWE-bench ProGPT-5.5 by 4.254.458.6+4.2
- SciCodeGPT-5.5 by 3.752.155.8+3.7
AgentsGPT-5.5
- MCP AtlasGPT-5.5 by 17.657.775.3+17.6
- GDPValGPT-5.5 by 16.925.842.7+16.9
- τ-Bench V3 · BankingGPT-5.5 by 13.425.639+13.4
- Terminal-Bench 4.0GPT-5.5 by 12.6214.6+12.6
- AA IT-Bench SREGPT-5.5 by 10.635.245.8+10.6
- AA ApexAgentsGPT-5.5 by 9.528.237.7+9.5
- OSWorld-VerifiedGPT-5.5 by 6.672.178.7+6.6
ReasoningGPT-5.5
- ARC-AGI-2GPT-5.5 by 66.118.985+66.1
- LiveBench · ReasoningGPT-5.5 by 18.471.389.7+18.4
- Humanity's Last ExamGPT-5.5 by 17.728.145.8+17.7
- CritPtGPT-5.5 by 17.11027.1+17.1
- GPQA DiamondGPT-5.5 by 687.593.5+6
FactsGPT-5.5
- SimpleQA VerifiedGPT-5.5 by 33.629.463+33.6
- AA-Omniscience · AccuracyGPT-5.5 by 20.537.558+20.5
- Vectara HHEM hallucination ratelower is betterGPT-5.4 mini by 3.85.59.3+3.8
- AA-Omniscience · Non-hallucinationGPT-5.5 by 1.29.811+1.2
Images and chartsGPT-5.5
- MMMU-ProGPT-5.5 by 6.673.379.9+6.6
- LMArena · VisionGPT-5.5 by 47 rating points12451292+47 rating
MathGPT-5.5
- FrontierMath Tier 4GPT-5.5 by 62.79.872.5+62.7
- FrontierMath Tiers 1-3 (v2)GPT-5.5 by 34.151.285.3+34.1
- LiveBench · MathematicsGPT-5.5 by 17.478.595.9+17.4
Following instructionsGPT-5.5
- LiveBench · Instruction FollowingGPT-5.5 by 10.959.870.7+10.9
- IFBenchGPT-5.5 by 2.673.375.9+2.6
Other results16 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-OmniscienceGPT-5.5 by 39.4-18.920.5+39.4
- frontiermath_tier_4_v1GPT-5.5 by 33.32.135.4+33.3
- ARC-AGI-1GPT-5.5 by 31.363.795+31.3
- Terminal-Bench 2.0GPT-5.5 by 22.76082.7+22.7
- Artificial Analysis Coding IndexGPT-5.5 by 18.856.174.9+18.8
- AA Agentic IndexGPT-5.5 by 17.619.737.3+17.6
- livebench_languageGPT-5.5 by 16.47187.4+16.4
- AA IntelligenceGPT-5.5 by 14.324.138.4+14.3
- ToolathlonGPT-5.5 by 12.742.955.6+12.7
- livebench_data_analysisGPT-5.5 by 10.870.881.6+10.8
- τ²-Bench Telecom (AA run)GPT-5.5 by 10.683.393.9+10.6
- vectara_avg_summary_lengthGPT-5.5 by 74.9 rating points54.7129.6+74.9 rating
- Finance Agent v2GPT-5.5 by 6.445.451.8+6.4
- vectara_factual_consistencyGPT-5.4 mini by 3.894.590.7+3.8
- Legal Agent BenchmarkGPT-5.5 by 2.102.1+2.1
- vectara_answer_ratetie100100tie
Questions people ask
Which is better, GPT-5.4 mini or GPT-5.5?
GPT-5.5 wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. GPT-5.4 mini wins none, but costs 85% less.
Which is better for coding?
GPT-5.5. It wins 7 of the 7 coding tests both models report; GPT-5.4 mini wins none.
Which is cheaper?
GPT-5.4 mini costs $0.75 per million input tokens and $4.50 per million output tokens; GPT-5.5 costs $5.00 and $30.00. That makes GPT-5.4 mini about 85% cheaper for the same work.
How do you compare the two?
We use the 47 benchmark tests both models have published scores on. The verdict counts the 31 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 16 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.