Model comparison
GPT-5.4 vs Grok-3 mini
GPT-5.4 is the stronger model overall, scoring 59.5 to 41.0 on the Noometry Index.
Last verified . 25 shared benchmarks.
Summary
- They share 25 benchmarks with published results for both. GPT-5.4 scores higher in 7 categories and Grok-3 mini in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.4 leads 61.7 to 13.8.
- The biggest single-benchmark swing is ARC-AGI-1: 93.7% for GPT-5.4 and 16.5% for Grok-3 mini.
Side by side
| GPT-5.4 | Grok-3 mini | |
|---|---|---|
| Provider | OpenAI | xAI |
| Noometry Index | 59.5 | 41.0 |
| Released | 2026-03-05 | 2025-04-09 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | — |
| Max output | 128K | — |
| Input $ / M tokens | $2.50 | — |
| Output $ / M tokens | $15 | — |
| Results tracked | 68 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 leads
GPT-5.4: 52.8 (#28), Grok-3 mini: 40.4 (#126)
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| WeirdML | 77.7% | 42.6% |
| LMArena Coding | 1497 | 1380 |
| SWE-bench Verified | 76.9% | — |
| DeepSWE | 51.8% | — |
| Aider Polyglot | — | 49.3% |
| LMArena WebDev | 1465 | — |
| SciCode | 56.6% | — |
| GSO | 31.4% | — |
| MirrorCode | 15.6% | — |
| ALE-Bench | 1,607 | — |
| AlgoTune | 1.85 | — |
Agentic & Tool Use Not comparable
GPT-5.4: 46.5 (#12), Grok-3 mini: —
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| Terminal-Bench | 81.8% | — |
| APEX-Agents | 52.4% | — |
| 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-5.4: 61.7 (#17), Grok-3 mini: 13.8 (#323)
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| ARC-AGI-2 | 74% | 0.4% |
| Kagi LLM Benchmark | 63.8% | 61.3% |
| ARC-AGI-1 | 93.7% | 16.5% |
| LMArena Hard Prompts | 1485 | 1374 |
| Epoch Capabilities Index | 156.92 | 140.35 |
| NYT Connections (extended) | 91.3% | — |
| CritPt | 23.4% | — |
| Chess Puzzles | 44% | — |
| EnigmaEval | 16% | — |
| Thematic Generalization | 80% | — |
| EBR-Bench | 25.4% | — |
| Mystery Game Puzzles | 37% | — |
| DTBench | 94.4% | — |
| LMCA | 52% | — |
| ForecastBench | 59.5 | — |
Math GPT-5.4 leads
GPT-5.4: 73.3 (#14), Grok-3 mini: 40.3 (#78)
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 97.8% | 77.8% |
| LMArena Math | 1488 | 1386 |
| FrontierMath (Feb 2025 set) | 47.6% | 5.9% |
| FrontierMath (Tiers 1-3) | 78.6% | — |
| FrontierMath Tier 4 | 49% | — |
| MathArena Final-Answer Competitions | 83.1% | — |
| ProofBench | 56% | — |
| Omni-MATH | — | 31.8% |
| MATH Level 5 | — | 90.9% |
| FrontierMath Tier 4 (v1) | 27.1% | — |
Knowledge GPT-5.4 leads
GPT-5.4: 65.3 (#13), Grok-3 mini: 46.3 (#75)
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| GPQA Diamond | 93.3% | 76.3% |
| LMArena Expert | 1507 | 1395 |
| Humanity's Last Exam | 36.2% | — |
| SimpleQA Verified | 45.1% | — |
| MMLU-Pro | — | 79.9% |
| Confabulations | — | 10.8% |
| Vectara Hallucination Rate | 7% | — |
| GPQA (HELM) | — | 67.5% |
Multimodal Not comparable
GPT-5.4: 43.8 (#19), Grok-3 mini: —
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| LMArena Vision | 1302 | — |
| Blueprint-Bench 2 | 27.1% | — |
| Furniture Assembly | 37.5% | — |
| LMArena Document | 1471 | — |
Multilingual GPT-5.4 leads
GPT-5.4: 56.4 (#22), Grok-3 mini: 48.3 (#141)
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| LMArena Non-English | 1465 | 1352 |
| LMArena Chinese | 1520 | 1387 |
| LMArena French | 1493 | 1357 |
| LMArena German | 1472 | 1350 |
| LMArena Japanese | 1484 | 1343 |
| LMArena Korean | 1449 | 1336 |
| LMArena Russian | 1480 | 1352 |
| LMArena Spanish | 1453 | 1379 |
Instruction Following Grok-3 mini leads
GPT-5.4: 77.3 (#25), Grok-3 mini: 78.7 (#8)
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| LMArena Instruction Following | 1469 | 1356 |
| IFEval | — | 95.1% |
Long Context GPT-5.4 leads
GPT-5.4: 50.3 (#8), Grok-3 mini: 41.0 (#142)
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| LMArena Longer Query | 1474 | 1372 |
| Fiction.LiveBench | — | 66.7% |
| CL-bench | 27.9% | — |
| CL-bench Life | 21.7% | — |
Writing & Preference GPT-5.4 leads
GPT-5.4: 72.1 (#18), Grok-3 mini: 52.7 (#163)
| Benchmark | GPT-5.4 | Grok-3 mini |
|---|---|---|
| LMArena Text | 1469 | 1370 |
| LMArena Creative Writing | 1439 | 1342 |
| LMArena Multi-Turn | 1482 | 1355 |
| Short-Story Creative Writing | — | 73.5% |
| EQ-Bench Creative Writing | 1840 | — |
| WildBench | — | 65.1% |
| EQ-Bench 4 | 1272 | — |
Frequently asked questions
Is GPT-5.4 better than Grok-3 mini?
GPT-5.4 is the stronger model overall, scoring 59.5 to 41.0 on the Noometry Index.
Is GPT-5.4 or Grok-3 mini better for coding?
GPT-5.4 scores higher on coding benchmarks: 52.8 versus 40.4 in the Noometry coding category.
How many benchmarks do GPT-5.4 and Grok-3 mini share?
25 benchmarks have published results for both models. GPT-5.4 has 68 scored results on Noometry and Grok-3 mini has 35.