Model comparison
Gemini 3 Pro vs Grok 4.20 (Non-Reasoning)
Gemini 3 Pro is the stronger model overall, scoring 54.8 to 48.6 on the Noometry Index.
Last verified . 38 shared benchmarks.
GoogleSummary
- They share 38 benchmarks with published results for both. Gemini 3 Pro scores higher in 9 categories and Grok 4.20 (Non-Reasoning) in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in multimodal, where Gemini 3 Pro leads 57.6 to 33.3.
- The biggest single-benchmark swing is ARC-AGI-2: 31.1% for Gemini 3 Pro and 65.1% for Grok 4.20 (Non-Reasoning).
Side by side
| Gemini 3 Pro | Grok 4.20 (Non-Reasoning) | |
|---|---|---|
| Provider | xAI | |
| Noometry Index | 54.8 | 48.6 |
| Released | 2025-11-18 | 2026-02-17 |
| Weights | Proprietary | Proprietary |
| Context window | — | 1M |
| Max output | — | 30K |
| Input $ / M tokens | — | $1.25 |
| Output $ / M tokens | — | $2.50 |
| Results tracked | 67 | 46 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemini 3 Pro leads
Gemini 3 Pro: 51.6 (#39), Grok 4.20 (Non-Reasoning): 42.1 (#112)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena WebDev | 1440 | 1375 |
| WeirdML | 69.9% | 52.3% |
| LMArena Coding | 1481 | 1459 |
| ALE-Bench | 1,177 | 1,150 |
| SWE-bench Verified | 72.9% | — |
| SWE-bench Verified (bash only) | 74.2% | — |
| SWE-bench Multilingual | 68.7% | — |
| GSO | 18.6% | — |
| AlgoTune | 1.83 | — |
Agentic & Tool Use Gemini 3 Pro leads
Gemini 3 Pro: 40.6 (#23), Grok 4.20 (Non-Reasoning): 34.4 (#46)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| Terminal-Bench | 69.4% | 57.3% |
| τ²-bench Banking | 18% | 18% |
| LMArena Search | 1207 | 1189 |
| Vending-Bench 2 | 5,478 | 4,663 |
| Berkeley Function Calling Leaderboard | 72.5% | — |
| GDPval | 40.3% | — |
| Remote Labor Index | 1.3% | — |
| τ²-bench Airline | 80.5% | — |
| τ²-bench Retail | 75.9% | — |
| τ²-bench Telecom | 91% | — |
| DeepResearch Bench | 46.3% | — |
| BALROG | 58.1% | — |
| METR Time Horizons | 71% | — |
Reasoning Too close to call
Gemini 3 Pro: 52.5 (#31), Grok 4.20 (Non-Reasoning): 52.3 (#32)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| ARC-AGI-2 | 31.1% | 65.1% |
| Kagi LLM Benchmark | 80.1% | 75% |
| NYT Connections (extended) | 94.4% | 85.4% |
| ARC-AGI-1 | 75% | 89.5% |
| Chess Puzzles | 31% | 24% |
| LMArena Hard Prompts | 1480 | 1451 |
| Epoch Capabilities Index | 152.92 | 151.98 |
| ForecastBench | 61.2 | 61.4 |
| SimpleBench | 76.4% | — |
| CritPt | 6.9% | — |
| EnigmaEval | 18.2% | — |
| Thematic Generalization | — | 63.8% |
| DTBench | — | 90.1% |
| LMCA | — | 38.7% |
Math Gemini 3 Pro leads
Gemini 3 Pro: 49.9 (#59), Grok 4.20 (Non-Reasoning): 48.2 (#65)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 91.4% | 92.2% |
| ProofBench | 20% | 14% |
| LMArena Math | 1476 | 1455 |
| FrontierMath (Tiers 1-3) | — | 44.9% |
| FrontierMath Tier 4 | — | 17.1% |
| MathArena Final-Answer Competitions | 67% | — |
| Omni-MATH | 55.5% | — |
| FrontierMath (Feb 2025 set) | 37.6% | — |
| FrontierMath Tier 4 (v1) | 18.8% | — |
Knowledge Gemini 3 Pro leads
Gemini 3 Pro: 64.4 (#16), Grok 4.20 (Non-Reasoning): 52.8 (#60)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| GPQA Diamond | 92.6% | 89.3% |
| LMArena Expert | 1475 | 1439 |
| Humanity's Last Exam | 37.5% | — |
| SimpleQA Verified | — | 30.2% |
| MMLU-Pro | 90.3% | — |
| Vectara Hallucination Rate | 13.6% | — |
| GPQA (HELM) | 80.3% | — |
Multimodal Gemini 3 Pro leads
Gemini 3 Pro: 57.6 (#2), Grok 4.20 (Non-Reasoning): 33.3 (#98)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Vision | 1305 | 1263 |
| LMArena Document | 1434 | 1416 |
| GeoBench | 84% | — |
| VPCT | 91% | — |
| Blueprint-Bench 2 | — | 0% |
Multilingual Gemini 3 Pro leads
Gemini 3 Pro: 56.9 (#16), Grok 4.20 (Non-Reasoning): 54.5 (#40)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Non-English | 1474 | 1441 |
| LMArena Chinese | 1523 | 1481 |
| LMArena French | 1492 | 1476 |
| LMArena German | 1515 | 1465 |
| LMArena Japanese | 1510 | 1449 |
| LMArena Korean | 1448 | 1417 |
| LMArena Russian | 1493 | 1458 |
| LMArena Spanish | 1470 | 1443 |
Instruction Following Gemini 3 Pro leads
Gemini 3 Pro: 76.3 (#45), Grok 4.20 (Non-Reasoning): 74.8 (#83)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Instruction Following | 1458 | 1420 |
| IFEval | 87.7% | — |
Long Context Grok 4.20 (Non-Reasoning) leads
Gemini 3 Pro: 44.0 (#79), Grok 4.20 (Non-Reasoning): 45.5 (#34)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| CL-bench | 15.8% | 22.2% |
| LMArena Longer Query | 1471 | 1437 |
| CL-bench Life | — | 11.9% |
Writing & Preference Too close to call
Gemini 3 Pro: 66.4 (#35), Grok 4.20 (Non-Reasoning): 65.7 (#44)
| Benchmark | Gemini 3 Pro | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Text | 1479 | 1451 |
| LMArena Creative Writing | 1482 | 1438 |
| EQ-Bench Creative Writing | 1525 | 1574 |
| LMArena Multi-Turn | 1484 | 1456 |
| WildBench | 85.9% | — |
Frequently asked questions
Is Gemini 3 Pro better than Grok 4.20 (Non-Reasoning)?
Gemini 3 Pro is the stronger model overall, scoring 54.8 to 48.6 on the Noometry Index.
Is Gemini 3 Pro or Grok 4.20 (Non-Reasoning) better for coding?
Gemini 3 Pro scores higher on coding benchmarks: 51.6 versus 42.1 in the Noometry coding category.
How many benchmarks do Gemini 3 Pro and Grok 4.20 (Non-Reasoning) share?
38 benchmarks have published results for both models. Gemini 3 Pro has 67 scored results on Noometry and Grok 4.20 (Non-Reasoning) has 46.