Model comparison
Gemini 3 Pro vs Kimi K3
Kimi K3 is the stronger model overall, scoring 59.5 to 54.8 on the Noometry Index.
Last verified . 35 shared benchmarks.
GoogleSummary
- They share 35 benchmarks with published results for both. Gemini 3 Pro scores higher in 3 categories and Kimi K3 in 7 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K3 leads 74.2 to 49.9.
- The biggest single-benchmark swing is ProofBench: 20% for Gemini 3 Pro and 87% for Kimi K3.
- Kimi K3 has downloadable open weights; the other is API-only.
Side by side
| Gemini 3 Pro | Kimi K3 | |
|---|---|---|
| Provider | Moonshot AI | |
| Noometry Index | 54.8 | 59.5 |
| Released | 2025-11-18 | 2026-07-16 |
| Weights | Proprietary | Open |
| Context window | — | 1.05M |
| Max output | — | 1.05M |
| Input $ / M tokens | — | $3 |
| Output $ / M tokens | — | $15 |
| Results tracked | 67 | 53 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K3 leads
Gemini 3 Pro: 51.6 (#39), Kimi K3: 61.0 (#10)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| LMArena WebDev | 1440 | 1654 |
| WeirdML | 69.9% | 82.6% |
| LMArena Coding | 1481 | 1508 |
| ALE-Bench | 1,177 | 1,524 |
| SWE-bench Verified | 72.9% | — |
| DeepSWE | — | 68.5% |
| FrontierCode | — | 44.2% |
| SWE-bench Verified (bash only) | 74.2% | — |
| SWE-bench Multilingual | 68.7% | — |
| FrontierSWE | — | 25.9% |
| SciCode | — | 59.5% |
| GSO | 18.6% | — |
| AlgoTune | 1.83 | — |
Agentic & Tool Use Kimi K3 leads
Gemini 3 Pro: 40.6 (#23), Kimi K3: 41.8 (#20)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| τ²-bench Banking | 18% | 37.1% |
| Vending-Bench 2 | 5,478 | 5,165 |
| Terminal-Bench | 69.4% | — |
| APEX-Agents | — | 50.6% |
| 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% | — |
| PostTrainBench | — | 32% |
| BALROG | 58.1% | — |
| GBAEval | — | 48.3% |
| GDP.pdf | — | 19% |
| LMArena Search | 1207 | — |
| METR Time Horizons | 71% | — |
Reasoning Kimi K3 leads
Gemini 3 Pro: 52.5 (#31), Kimi K3: 63.0 (#17)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| ARC-AGI-2 | 31.1% | 60.4% |
| SimpleBench | 76.4% | 60.7% |
| NYT Connections (extended) | 94.4% | 93.6% |
| ARC-AGI-1 | 75% | 94.5% |
| CritPt | 6.9% | 23.4% |
| Chess Puzzles | 31% | 39% |
| LMArena Hard Prompts | 1480 | 1496 |
| Epoch Capabilities Index | 152.92 | 157.45 |
| ForecastBench | 61.2 | 61.1 |
| Kagi LLM Benchmark | 80.1% | — |
| EnigmaEval | 18.2% | — |
| Mystery Game Puzzles | — | 26% |
| DTBench | — | 91.2% |
| LMCA | — | 52.7% |
| Surface Evolver Bench | — | 95% |
Math Kimi K3 leads
Gemini 3 Pro: 49.9 (#59), Kimi K3: 74.2 (#16)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| MathArena Final-Answer Competitions | 67% | 87.8% |
| OTIS Mock AIME 2024-2025 | 91.4% | 97.2% |
| ProofBench | 20% | 87% |
| LMArena Math | 1476 | 1491 |
| FrontierMath (Tiers 1-3) | — | 72.2% |
| FrontierMath Tier 4 | — | 39% |
| 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), Kimi K3: 63.2 (#21)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| GPQA Diamond | 92.6% | 93.1% |
| LMArena Expert | 1475 | 1521 |
| Humanity's Last Exam | 37.5% | — |
| SimpleQA Verified | — | 50.6% |
| MMLU-Pro | 90.3% | — |
| Vectara Hallucination Rate | 13.6% | — |
| GPQA (HELM) | 80.3% | — |
Multimodal Gemini 3 Pro leads
Gemini 3 Pro: 57.6 (#2), Kimi K3: 37.8 (#70)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| LMArena Vision | 1305 | — |
| GeoBench | 84% | — |
| VPCT | 91% | — |
| Blueprint-Bench 2 | — | 29.5% |
| Furniture Assembly | — | 34.2% |
| LMArena Document | 1434 | — |
Multilingual Too close to call
Gemini 3 Pro: 56.9 (#16), Kimi K3: 56.3 (#21)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| LMArena Non-English | 1474 | 1466 |
| LMArena Chinese | 1523 | 1529 |
| LMArena French | 1492 | 1491 |
| LMArena German | 1515 | 1488 |
| LMArena Japanese | 1510 | 1487 |
| LMArena Korean | 1448 | 1458 |
| LMArena Russian | 1493 | 1482 |
| LMArena Spanish | 1470 | 1472 |
Instruction Following Kimi K3 leads
Gemini 3 Pro: 76.3 (#45), Kimi K3: 77.7 (#14)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| LMArena Instruction Following | 1458 | 1483 |
| IFEval | 87.7% | — |
Long Context Kimi K3 leads
Gemini 3 Pro: 44.0 (#79), Kimi K3: 45.8 (#29)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| LMArena Longer Query | 1471 | 1494 |
| CL-bench | 15.8% | — |
Writing & Preference Kimi K3 leads
Gemini 3 Pro: 66.4 (#35), Kimi K3: 76.6 (#4)
| Benchmark | Gemini 3 Pro | Kimi K3 |
|---|---|---|
| LMArena Text | 1479 | 1476 |
| LMArena Creative Writing | 1482 | 1454 |
| EQ-Bench Creative Writing | 1525 | 2082 |
| LMArena Multi-Turn | 1484 | 1488 |
| WildBench | 85.9% | — |
| EQ-Bench 4 | — | 1339 |
Frequently asked questions
Is Gemini 3 Pro better than Kimi K3?
Kimi K3 is the stronger model overall, scoring 59.5 to 54.8 on the Noometry Index.
Is Gemini 3 Pro or Kimi K3 better for coding?
Kimi K3 scores higher on coding benchmarks: 61.0 versus 51.6 in the Noometry coding category.
How many benchmarks do Gemini 3 Pro and Kimi K3 share?
35 benchmarks have published results for both models. Gemini 3 Pro has 67 scored results on Noometry and Kimi K3 has 53.