Model comparison
GPT-5.6 Luna vs Grok 4.20 (Non-Reasoning)
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 48.6 on the Noometry Index.
Last verified . 39 shared benchmarks.
Summary
- They share 39 benchmarks with published results for both. GPT-5.6 Luna scores higher in 6 categories and Grok 4.20 (Non-Reasoning) in 4 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Luna leads 77.7 to 48.2.
- The biggest single-benchmark swing is ProofBench: 60% for GPT-5.6 Luna and 14% for Grok 4.20 (Non-Reasoning).
- GPT-5.6 Luna is cheaper at $0.20 / $1.20 per million input/output tokens, against $1.25 / $2.50 for Grok 4.20 (Non-Reasoning).
- GPT-5.6 Luna accepts more context: 1.05M tokens versus 1M.
Side by side
| GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) | |
|---|---|---|
| Provider | OpenAI | xAI |
| Noometry Index | 54.6 | 48.6 |
| Released | 2026-07-09 | 2026-02-17 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 1M |
| Max output | 128K | 30K |
| Input $ / M tokens | $0.20 | $1.25 |
| Output $ / M tokens | $1.20 | $2.50 |
| Results tracked | 52 | 46 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.6 Luna leads
GPT-5.6 Luna: 54.5 (#28), Grok 4.20 (Non-Reasoning): 42.1 (#112)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena WebDev | 1519 | 1375 |
| WeirdML | 60.9% | 52.3% |
| LMArena Coding | 1466 | 1459 |
| ALE-Bench | 1,667 | 1,150 |
| DeepSWE | 67.2% | — |
| FrontierCode | 39.8% | — |
| CursorBench | 35.9% | — |
| SciCode | 53.6% | — |
Agentic & Tool Use Too close to call
GPT-5.6 Luna: 34.4 (#45), Grok 4.20 (Non-Reasoning): 34.4 (#46)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| Vending-Bench 2 | 4,095 | 4,663 |
| Terminal-Bench | — | 57.3% |
| APEX-Agents | 43% | — |
| τ²-bench Banking | — | 18% |
| BALROG | 45.6% | — |
| GDP.pdf | 22.7% | — |
| LMArena Search | — | 1189 |
Reasoning Grok 4.20 (Non-Reasoning) leads
GPT-5.6 Luna: 47.6 (#43), Grok 4.20 (Non-Reasoning): 52.3 (#32)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| ARC-AGI-2 | 59.5% | 65.1% |
| Kagi LLM Benchmark | 49.1% | 75% |
| NYT Connections (extended) | 69.4% | 85.4% |
| ARC-AGI-1 | 88% | 89.5% |
| Chess Puzzles | 40% | 24% |
| LMArena Hard Prompts | 1451 | 1451 |
| DTBench | 89.1% | 90.1% |
| LMCA | 48.5% | 38.7% |
| Epoch Capabilities Index | 156.39 | 151.98 |
| SimpleBench | 46.8% | — |
| CritPt | 20.6% | — |
| Thematic Generalization | — | 63.8% |
| Mystery Game Puzzles | 21% | — |
| Surface Evolver Bench | 61.9% | — |
| ForecastBench | — | 61.4 |
Math GPT-5.6 Luna leads
GPT-5.6 Luna: 77.7 (#14), Grok 4.20 (Non-Reasoning): 48.2 (#65)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| FrontierMath (Tiers 1-3) | 82.1% | 44.9% |
| FrontierMath Tier 4 | 61% | 17.1% |
| OTIS Mock AIME 2024-2025 | 98.3% | 92.2% |
| ProofBench | 60% | 14% |
| LMArena Math | 1458 | 1455 |
Knowledge GPT-5.6 Luna leads
GPT-5.6 Luna: 58.5 (#34), Grok 4.20 (Non-Reasoning): 52.8 (#60)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| GPQA Diamond | 91.6% | 89.3% |
| SimpleQA Verified | 41% | 30.2% |
| LMArena Expert | 1478 | 1439 |
Multimodal GPT-5.6 Luna leads
GPT-5.6 Luna: 42.7 (#28), Grok 4.20 (Non-Reasoning): 33.3 (#98)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Vision | 1258 | 1263 |
| Blueprint-Bench 2 | 22.6% | 0% |
| LMArena Document | 1457 | 1416 |
| Furniture Assembly | 42.5% | — |
Multilingual Grok 4.20 (Non-Reasoning) leads
GPT-5.6 Luna: 52.8 (#78), Grok 4.20 (Non-Reasoning): 54.5 (#40)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Non-English | 1417 | 1441 |
| LMArena Chinese | 1470 | 1481 |
| LMArena French | 1456 | 1476 |
| LMArena German | 1454 | 1465 |
| LMArena Japanese | 1411 | 1449 |
| LMArena Korean | 1415 | 1417 |
| LMArena Russian | 1428 | 1458 |
| LMArena Spanish | 1448 | 1443 |
Instruction Following Too close to call
GPT-5.6 Luna: 75.6 (#57), Grok 4.20 (Non-Reasoning): 74.8 (#83)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Instruction Following | 1437 | 1420 |
Long Context Grok 4.20 (Non-Reasoning) leads
GPT-5.6 Luna: 43.9 (#82), Grok 4.20 (Non-Reasoning): 45.5 (#34)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Longer Query | 1436 | 1437 |
| CL-bench | — | 22.2% |
| CL-bench Life | — | 11.9% |
Writing & Preference GPT-5.6 Luna leads
GPT-5.6 Luna: 68.0 (#29), Grok 4.20 (Non-Reasoning): 65.7 (#44)
| Benchmark | GPT-5.6 Luna | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Text | 1431 | 1451 |
| LMArena Creative Writing | 1396 | 1438 |
| EQ-Bench Creative Writing | 1829 | 1574 |
| LMArena Multi-Turn | 1434 | 1456 |
| EQ-Bench 4 | 1156 | — |
Frequently asked questions
Is GPT-5.6 Luna better than Grok 4.20 (Non-Reasoning)?
GPT-5.6 Luna is the stronger model overall, scoring 54.6 to 48.6 on the Noometry Index.
Which is cheaper, GPT-5.6 Luna or Grok 4.20 (Non-Reasoning)?
GPT-5.6 Luna is cheaper. It lists at $0.20 per million input tokens and $1.20 per million output tokens; Grok 4.20 (Non-Reasoning) lists at $1.25 and $2.50.
Is GPT-5.6 Luna or Grok 4.20 (Non-Reasoning) better for coding?
GPT-5.6 Luna scores higher on coding benchmarks: 54.5 versus 42.1 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Luna does, with 1.05M tokens against 1M.
How many benchmarks do GPT-5.6 Luna and Grok 4.20 (Non-Reasoning) share?
39 benchmarks have published results for both models. GPT-5.6 Luna has 52 scored results on Noometry and Grok 4.20 (Non-Reasoning) has 46.