Model comparison
GPT-4.1 mini vs GPT-5.4
GPT-5.4 is the stronger model overall, scoring 59.4 to 33.6 on the Noometry Index. GPT-4.1 mini costs 8.0× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.
Last verified . 35 shared benchmarks.
Summary
- They share 35 benchmarks with published results for both. GPT-4.1 mini scores higher in 0 categories and GPT-5.4 in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.4 leads 61.8 to 10.8.
- The biggest single-benchmark swing is ARC-AGI-1: 3.5% for GPT-4.1 mini and 93.7% for GPT-5.4.
- GPT-4.1 mini is cheaper at $0.40 / $1.60 per million input/output tokens, against $2.50 / $15 for GPT-5.4.
- GPT-5.4 accepts more context: 1.05M tokens versus 1.05M.
Side by side
| GPT-4.1 mini | GPT-5.4 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 33.6 | 59.4 |
| Released | 2025-04-14 | 2026-03-05 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 1.05M |
| Max output | 33K | 128K |
| Input $ / M tokens | $0.40 | $2.50 |
| Output $ / M tokens | $1.60 | $15 |
| Results tracked | 47 | 68 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 leads
GPT-4.1 mini: 30.6 (#293), GPT-5.4: 52.6 (#33)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| SciCode | 40.4% | 56.6% |
| WeirdML | 37.6% | 77.7% |
| LMArena Coding | 1367 | 1497 |
| SWE-bench Verified | — | 76.9% |
| DeepSWE | — | 51.8% |
| SWE-bench Verified (bash only) | 23.9% | — |
| Aider Polyglot | 32.4% | — |
| LMArena WebDev | — | 1465 |
| GSO | — | 31.4% |
| BigCodeBench Instruct | 48.9% | — |
| MirrorCode | — | 15.6% |
| CadEval | 16% | — |
| ALE-Bench | — | 1,607 |
| AlgoTune | — | 1.85 |
Agentic & Tool Use GPT-5.4 leads
GPT-4.1 mini: 33.3 (#55), GPT-5.4: 46.5 (#13)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| Terminal-Bench | — | 81.8% |
| APEX-Agents | — | 52.4% |
| Berkeley Function Calling Leaderboard | 50.5% | — |
| τ²-bench Banking | — | 39.4% |
| DeepResearch Bench | — | 35.1% |
| PostTrainBench | — | 19% |
| GBAEval | — | 45.1% |
| LMArena Search | — | 1197 |
| METR Time Horizons | — | 74.3% |
| Vending-Bench 2 | — | 6,144 |
Reasoning GPT-5.4 leads
GPT-4.1 mini: 10.8 (#340), GPT-5.4: 61.8 (#19)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| ARC-AGI-2 | 0% | 74% |
| Kagi LLM Benchmark | 48.6% | 63.8% |
| ARC-AGI-1 | 3.5% | 93.7% |
| CritPt | 0% | 23.4% |
| Chess Puzzles | 7% | 44% |
| LMArena Hard Prompts | 1349 | 1485 |
| Mystery Game Puzzles | 7% | 37% |
| DTBench | 68.8% | 94.4% |
| LMCA | 21.1% | 52% |
| Epoch Capabilities Index | 135.01 | 156.81 |
| NYT Connections (extended) | — | 91.3% |
| EnigmaEval | — | 16% |
| Thematic Generalization | — | 80% |
| EBR-Bench | — | 25.4% |
| ForecastBench | — | 59.5 |
Math GPT-5.4 leads
GPT-4.1 mini: 24.1 (#270), GPT-5.4: 73.5 (#19)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 6.7% | 78.6% |
| OTIS Mock AIME 2024-2025 | 44.7% | 97.8% |
| LMArena Math | 1343 | 1488 |
| FrontierMath (Feb 2025 set) | 4.5% | 47.6% |
| FrontierMath Tier 4 | — | 49% |
| MathArena Final-Answer Competitions | — | 83.1% |
| ProofBench | — | 56% |
| Omni-MATH | 49.1% | — |
| MATH Level 5 | 87.3% | — |
| FrontierMath Tier 4 (v1) | — | 27.1% |
Knowledge GPT-5.4 leads
GPT-4.1 mini: 34.7 (#194), GPT-5.4: 65.3 (#14)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| GPQA Diamond | 65.8% | 93.3% |
| SimpleQA Verified | 12.7% | 45.1% |
| LMArena Expert | 1338 | 1507 |
| Humanity's Last Exam | — | 36.2% |
| MMLU-Pro | 78.3% | — |
| Vectara Hallucination Rate | — | 7% |
| GPQA (HELM) | 61.4% | — |
Multimodal GPT-5.4 leads
GPT-4.1 mini: 35.8 (#82), GPT-5.4: 43.7 (#20)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| LMArena Vision | 1181 | 1303 |
| Blueprint-Bench 2 | — | 27.1% |
| Furniture Assembly | — | 37.5% |
| LMArena Document | — | 1471 |
Multilingual GPT-5.4 leads
GPT-4.1 mini: 45.7 (#166), GPT-5.4: 56.2 (#23)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| LMArena Non-English | 1318 | 1465 |
| LMArena Chinese | 1329 | 1519 |
| LMArena French | 1358 | 1493 |
| LMArena German | 1351 | 1472 |
| LMArena Japanese | 1290 | 1485 |
| LMArena Korean | 1298 | 1448 |
| LMArena Russian | 1324 | 1480 |
| LMArena Spanish | 1319 | 1454 |
Instruction Following GPT-5.4 leads
GPT-4.1 mini: 73.7 (#118), GPT-5.4: 77.1 (#27)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| LMArena Instruction Following | 1333 | 1469 |
| IFEval | 90.4% | — |
Long Context GPT-5.4 leads
GPT-4.1 mini: 31.8 (#275), GPT-5.4: 50.3 (#8)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| LMArena Longer Query | 1344 | 1473 |
| Fiction.LiveBench | 44.4% | — |
| CL-bench | — | 27.9% |
| CL-bench Life | — | 21.7% |
Writing & Preference GPT-5.4 leads
GPT-4.1 mini: 48.6 (#199), GPT-5.4: 71.9 (#17)
| Benchmark | GPT-4.1 mini | GPT-5.4 |
|---|---|---|
| LMArena Text | 1340 | 1469 |
| LMArena Creative Writing | 1300 | 1439 |
| EQ-Bench Creative Writing | 1147 | 1840 |
| LMArena Multi-Turn | 1354 | 1482 |
| WildBench | 83.8% | — |
| EQ-Bench 4 | — | 1272 |
Frequently asked questions
Is GPT-4.1 mini better than GPT-5.4?
GPT-5.4 is the stronger model overall, scoring 59.4 to 33.6 on the Noometry Index. GPT-4.1 mini costs 8.0× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.
Which is cheaper, GPT-4.1 mini or GPT-5.4?
GPT-4.1 mini is cheaper. It lists at $0.40 per million input tokens and $1.60 per million output tokens; GPT-5.4 lists at $2.50 and $15.
Is GPT-4.1 mini or GPT-5.4 better for coding?
GPT-5.4 scores higher on coding benchmarks: 52.6 versus 30.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 does, with 1.05M tokens against 1.05M.
How many benchmarks do GPT-4.1 mini and GPT-5.4 share?
35 benchmarks have published results for both models. GPT-4.1 mini has 47 scored results on Noometry and GPT-5.4 has 68.