Model comparison
GPT-5.4 vs Grok 4.6
GPT-5.4 is the stronger model overall, scoring 59.4 to 56.9 on the Noometry Index. Grok 4.6 costs 1.9× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.
Last verified . 44 shared benchmarks.
Summary
- They share 44 benchmarks with published results for both. GPT-5.4 scores higher in 9 categories and Grok 4.6 in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where GPT-5.4 leads 71.9 to 62.3.
- The biggest single-benchmark swing is FrontierMath Tier 4: 49% for GPT-5.4 and 31.7% for Grok 4.6.
- Grok 4.6 is cheaper at $2 / $6 per million input/output tokens, against $2.50 / $15 for GPT-5.4.
- GPT-5.4 accepts more context: 1.05M tokens versus 500K.
Side by side
| GPT-5.4 | Grok 4.6 | |
|---|---|---|
| Provider | OpenAI | xAI |
| Noometry Index | 59.4 | 56.9 |
| Released | 2026-03-05 | 2026-08-12 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 500K |
| Max output | 128K | 500K |
| Input $ / M tokens | $2.50 | $2 |
| Output $ / M tokens | $15 | $6 |
| Results tracked | 68 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4.6 leads
GPT-5.4: 52.6 (#33), Grok 4.6: 58.5 (#16)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| DeepSWE | 51.8% | 67.5% |
| LMArena WebDev | 1465 | 1617 |
| SciCode | 56.6% | 56.5% |
| WeirdML | 77.7% | 67.3% |
| LMArena Coding | 1497 | 1465 |
| ALE-Bench | 1,607 | 1,508 |
| SWE-bench Verified | 76.9% | — |
| FrontierCode | — | 48% |
| CursorBench | — | 41.4% |
| FrontierSWE | — | 25.3% |
| GSO | 31.4% | — |
| MirrorCode | 15.6% | — |
| AlgoTune | 1.85 | — |
Agentic & Tool Use GPT-5.4 leads
GPT-5.4: 46.5 (#13), Grok 4.6: 39.4 (#27)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| APEX-Agents | 52.4% | 65.3% |
| Vending-Bench 2 | 6,144 | 9,047 |
| Terminal-Bench | 81.8% | — |
| τ²-bench Banking | 39.4% | — |
| DeepResearch Bench | 35.1% | — |
| PostTrainBench | 19% | — |
| GBAEval | 45.1% | — |
| GDP.pdf | — | 17.2% |
| LMArena Search | 1197 | — |
| METR Time Horizons | 74.3% | — |
Reasoning Too close to call
GPT-5.4: 61.8 (#19), Grok 4.6: 61.4 (#20)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| ARC-AGI-2 | 74% | 67.1% |
| NYT Connections (extended) | 91.3% | 80% |
| ARC-AGI-1 | 93.7% | 87.5% |
| CritPt | 23.4% | 19.7% |
| Chess Puzzles | 44% | 40% |
| EBR-Bench | 25.4% | 30.5% |
| LMArena Hard Prompts | 1485 | 1447 |
| Mystery Game Puzzles | 37% | 34% |
| DTBench | 94.4% | 97.3% |
| LMCA | 52% | 48.5% |
| Epoch Capabilities Index | 156.81 | 156.44 |
| SimpleBench | — | 75.9% |
| Kagi LLM Benchmark | 63.8% | — |
| EnigmaEval | 16% | — |
| Thematic Generalization | 80% | — |
| ForecastBench | 59.5 | — |
Math GPT-5.4 leads
GPT-5.4: 73.5 (#19), Grok 4.6: 67.0 (#24)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 78.6% | 66% |
| FrontierMath Tier 4 | 49% | 31.7% |
| OTIS Mock AIME 2024-2025 | 97.8% | 99.2% |
| ProofBench | 56% | 51% |
| LMArena Math | 1488 | 1423 |
| MathArena Final-Answer Competitions | 83.1% | — |
| FrontierMath (Feb 2025 set) | 47.6% | — |
| FrontierMath Tier 4 (v1) | 27.1% | — |
Knowledge GPT-5.4 leads
GPT-5.4: 65.3 (#14), Grok 4.6: 63.3 (#20)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| GPQA Diamond | 93.3% | 94% |
| SimpleQA Verified | 45.1% | 49.3% |
| LMArena Expert | 1507 | 1467 |
| Humanity's Last Exam | 36.2% | — |
| Vectara Hallucination Rate | 7% | — |
Multimodal Too close to call
GPT-5.4: 43.7 (#20), Grok 4.6: 43.6 (#23)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| LMArena Vision | 1303 | 1263 |
| Blueprint-Bench 2 | 27.1% | 33.2% |
| Furniture Assembly | 37.5% | 40% |
| LMArena Document | 1471 | 1452 |
Multilingual GPT-5.4 leads
GPT-5.4: 56.2 (#23), Grok 4.6: 53.0 (#74)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| LMArena Non-English | 1465 | 1420 |
| LMArena Chinese | 1519 | 1480 |
| LMArena French | 1493 | 1461 |
| LMArena German | 1472 | 1431 |
| LMArena Japanese | 1485 | 1376 |
| LMArena Korean | 1448 | 1397 |
| LMArena Russian | 1480 | 1422 |
| LMArena Spanish | 1454 | 1404 |
Instruction Following GPT-5.4 leads
GPT-5.4: 77.1 (#27), Grok 4.6: 75.4 (#63)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| LMArena Instruction Following | 1469 | 1431 |
Long Context GPT-5.4 leads
GPT-5.4: 50.3 (#8), Grok 4.6: 44.5 (#66)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| LMArena Longer Query | 1473 | 1454 |
| CL-bench | 27.9% | — |
| CL-bench Life | 21.7% | — |
Writing & Preference GPT-5.4 leads
GPT-5.4: 71.9 (#17), Grok 4.6: 62.3 (#80)
| Benchmark | GPT-5.4 | Grok 4.6 |
|---|---|---|
| LMArena Text | 1469 | 1428 |
| LMArena Creative Writing | 1439 | 1428 |
| LMArena Multi-Turn | 1482 | 1425 |
| EQ-Bench Creative Writing | 1840 | — |
| EQ-Bench 4 | 1272 | — |
Frequently asked questions
Is GPT-5.4 better than Grok 4.6?
GPT-5.4 is the stronger model overall, scoring 59.4 to 56.9 on the Noometry Index. Grok 4.6 costs 1.9× 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-5.4 or Grok 4.6?
Grok 4.6 is cheaper. It lists at $2 per million input tokens and $6 per million output tokens; GPT-5.4 lists at $2.50 and $15.
Is GPT-5.4 or Grok 4.6 better for coding?
Grok 4.6 scores higher on coding benchmarks: 58.5 versus 52.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 does, with 1.05M tokens against 500K.
How many benchmarks do GPT-5.4 and Grok 4.6 share?
44 benchmarks have published results for both models. GPT-5.4 has 68 scored results on Noometry and Grok 4.6 has 49.