Model comparison
Grok 4.5 vs Kimi K3
Kimi K3 is the stronger model overall, scoring 59.5 to 55.0 on the Noometry Index. Grok 4.5 costs 2.0× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Last verified . 48 shared benchmarks.
Summary
- They share 48 benchmarks with published results for both. Grok 4.5 scores higher in 1 category and Kimi K3 in 9 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K3 leads 74.2 to 60.9.
- The biggest single-benchmark swing is ProofBench: 31% for Grok 4.5 and 87% for Kimi K3.
- Grok 4.5 is cheaper at $2 / $6 per million input/output tokens, against $3 / $15 for Kimi K3.
- Kimi K3 accepts more context: 1.05M tokens versus 500K.
- Kimi K3 has downloadable open weights; the other is API-only.
Side by side
| Grok 4.5 | Kimi K3 | |
|---|---|---|
| Provider | xAI | Moonshot AI |
| Noometry Index | 55.0 | 59.5 |
| Released | 2026-07-08 | 2026-07-16 |
| Weights | Proprietary | Open |
| Context window | 500K | 1.05M |
| Max output | 500K | 1.05M |
| Input $ / M tokens | $2 | $3 |
| Output $ / M tokens | $6 | $15 |
| Results tracked | 52 | 53 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K3 leads
Grok 4.5: 52.2 (#35), Kimi K3: 61.0 (#10)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| DeepSWE | 53.8% | 68.5% |
| FrontierCode | 42.4% | 44.2% |
| LMArena WebDev | 1553 | 1654 |
| SciCode | 54.1% | 59.5% |
| WeirdML | 46.4% | 82.6% |
| LMArena Coding | 1474 | 1508 |
| ALE-Bench | 1,309 | 1,524 |
| FrontierSWE | — | 25.9% |
Agentic & Tool Use Grok 4.5 leads
Grok 4.5: 44.4 (#17), Kimi K3: 41.8 (#20)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| APEX-Agents | 56.2% | 50.6% |
| τ²-bench Banking | 47.9% | 37.1% |
| PostTrainBench | 23.4% | 32% |
| GBAEval | 65.4% | 48.3% |
| GDP.pdf | 14% | 19% |
| Vending-Bench 2 | 3,887 | 5,165 |
| LMArena Search | 1213 | — |
Reasoning Kimi K3 leads
Grok 4.5: 56.1 (#25), Kimi K3: 63.0 (#17)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| ARC-AGI-2 | 52.6% | 60.4% |
| SimpleBench | 70% | 60.7% |
| NYT Connections (extended) | 79.9% | 93.6% |
| ARC-AGI-1 | 87.2% | 94.5% |
| CritPt | 15.4% | 23.4% |
| Chess Puzzles | 36% | 39% |
| LMArena Hard Prompts | 1462 | 1496 |
| DTBench | 96.5% | 91.2% |
| LMCA | 45.2% | 52.7% |
| Surface Evolver Bench | 74.4% | 95% |
| Epoch Capabilities Index | 153.92 | 157.45 |
| Kagi LLM Benchmark | 83.5% | — |
| Mystery Game Puzzles | — | 26% |
| ForecastBench | — | 61.1 |
Math Kimi K3 leads
Grok 4.5: 60.9 (#35), Kimi K3: 74.2 (#16)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 57.2% | 72.2% |
| FrontierMath Tier 4 | 24.4% | 39% |
| OTIS Mock AIME 2024-2025 | 97.8% | 97.2% |
| ProofBench | 31% | 87% |
| LMArena Math | 1459 | 1491 |
| MathArena Final-Answer Competitions | — | 87.8% |
Knowledge Too close to call
Grok 4.5: 62.3 (#24), Kimi K3: 63.2 (#21)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| GPQA Diamond | 93.4% | 93.1% |
| SimpleQA Verified | 48.3% | 50.6% |
| LMArena Expert | 1466 | 1521 |
Multimodal Too close to call
Grok 4.5: 37.6 (#72), Kimi K3: 37.8 (#70)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| Blueprint-Bench 2 | 27.3% | 29.5% |
| Furniture Assembly | 22.5% | 34.2% |
| LMArena Vision | 1288 | — |
| LMArena Document | 1452 | — |
Multilingual Kimi K3 leads
Grok 4.5: 54.4 (#42), Kimi K3: 56.3 (#21)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| LMArena Non-English | 1440 | 1466 |
| LMArena Chinese | 1496 | 1529 |
| LMArena French | 1456 | 1491 |
| LMArena German | 1446 | 1488 |
| LMArena Japanese | 1428 | 1487 |
| LMArena Korean | 1404 | 1458 |
| LMArena Russian | 1448 | 1482 |
| LMArena Spanish | 1450 | 1472 |
Instruction Following Kimi K3 leads
Grok 4.5: 76.0 (#48), Kimi K3: 77.7 (#14)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| LMArena Instruction Following | 1446 | 1483 |
Long Context Kimi K3 leads
Grok 4.5: 44.8 (#56), Kimi K3: 45.8 (#29)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| LMArena Longer Query | 1463 | 1494 |
Writing & Preference Kimi K3 leads
Grok 4.5: 65.8 (#42), Kimi K3: 76.6 (#4)
| Benchmark | Grok 4.5 | Kimi K3 |
|---|---|---|
| LMArena Text | 1448 | 1476 |
| LMArena Creative Writing | 1442 | 1454 |
| EQ-Bench Creative Writing | 1579 | 2082 |
| LMArena Multi-Turn | 1456 | 1488 |
| EQ-Bench 4 | — | 1339 |
Frequently asked questions
Is Grok 4.5 better than Kimi K3?
Kimi K3 is the stronger model overall, scoring 59.5 to 55.0 on the Noometry Index. Grok 4.5 costs 2.0× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Which is cheaper, Grok 4.5 or Kimi K3?
Grok 4.5 is cheaper. It lists at $2 per million input tokens and $6 per million output tokens; Kimi K3 lists at $3 and $15.
Is Grok 4.5 or Kimi K3 better for coding?
Kimi K3 scores higher on coding benchmarks: 61.0 versus 52.2 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 500K.
How many benchmarks do Grok 4.5 and Kimi K3 share?
48 benchmarks have published results for both models. Grok 4.5 has 52 scored results on Noometry and Kimi K3 has 53.