Model comparison
Kimi K3 vs Llama 4 Maverick
Kimi K3 is the stronger model overall, scoring 59.5 to 30.9 on the Noometry Index. Llama 4 Maverick costs 20× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Last verified . 32 shared benchmarks.
MetaSummary
- They share 32 benchmarks with published results for both. Kimi K3 scores higher in 10 categories and Llama 4 Maverick in 0 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Kimi K3 leads 63.0 to 10.1.
- The biggest single-benchmark swing is ARC-AGI-1: 94.5% for Kimi K3 and 4.4% for Llama 4 Maverick.
- Llama 4 Maverick is cheaper at $0.19 / $0.65 per million input/output tokens, against $3 / $15 for Kimi K3.
- Kimi K3 accepts more context: 1.05M tokens versus 128K.
Side by side
| Kimi K3 | Llama 4 Maverick | |
|---|---|---|
| Provider | Moonshot AI | Meta |
| Noometry Index | 59.5 | 30.9 |
| Released | 2026-07-16 | 2025-04-05 |
| Weights | Open | Open |
| Context window | 1.05M | 128K |
| Max output | 1.05M | 4K |
| Input $ / M tokens | $3 | $0.19 |
| Output $ / M tokens | $15 | $0.65 |
| Results tracked | 53 | 54 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K3 leads
Kimi K3: 61.0 (#10), Llama 4 Maverick: 26.6 (#324)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| SciCode | 59.5% | 33.1% |
| WeirdML | 82.6% | 24.5% |
| LMArena Coding | 1508 | 1302 |
| ALE-Bench | 1,524 | 172.97 |
| DeepSWE | 68.5% | — |
| FrontierCode | 44.2% | — |
| SWE-bench Verified (bash only) | — | 21% |
| Aider Polyglot | — | 15.6% |
| LMArena WebDev | 1654 | — |
| FrontierSWE | 25.9% | — |
| BigCodeBench Instruct | — | 49.7% |
| BigCodeBench Complete | — | 61.4% |
Agentic & Tool Use Kimi K3 leads
Kimi K3: 41.8 (#20), Llama 4 Maverick: 28.2 (#91)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| APEX-Agents | 50.6% | — |
| Berkeley Function Calling Leaderboard | — | 37.3% |
| τ²-bench Banking | 37.1% | — |
| PostTrainBench | 32% | — |
| GBAEval | 48.3% | — |
| GDP.pdf | 19% | — |
| Vending-Bench 2 | 5,165 | — |
Reasoning Kimi K3 leads
Kimi K3: 63.0 (#17), Llama 4 Maverick: 10.1 (#342)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| ARC-AGI-2 | 60.4% | 0% |
| SimpleBench | 60.7% | 27.7% |
| NYT Connections (extended) | 93.6% | 8% |
| ARC-AGI-1 | 94.5% | 4.4% |
| CritPt | 23.4% | 0% |
| LMArena Hard Prompts | 1496 | 1281 |
| DTBench | 91.2% | 61.9% |
| LMCA | 52.7% | 15.9% |
| Epoch Capabilities Index | 157.45 | 132.2 |
| ForecastBench | 61.1 | 57.5 |
| Kagi LLM Benchmark | — | 55.9% |
| Chess Puzzles | 39% | — |
| EnigmaEval | — | 0.6% |
| Mystery Game Puzzles | 26% | — |
| Surface Evolver Bench | 95% | — |
Math Kimi K3 leads
Kimi K3: 74.2 (#16), Llama 4 Maverick: 26.0 (#262)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 97.2% | 20.6% |
| LMArena Math | 1491 | 1299 |
| FrontierMath (Tiers 1-3) | 72.2% | — |
| FrontierMath Tier 4 | 39% | — |
| MathArena Final-Answer Competitions | 87.8% | — |
| ProofBench | 87% | — |
| Omni-MATH | — | 42.2% |
| MATH Level 5 | — | 73% |
| FrontierMath (Feb 2025 set) | — | 0.7% |
Knowledge Kimi K3 leads
Kimi K3: 63.2 (#21), Llama 4 Maverick: 33.4 (#204)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| GPQA Diamond | 93.1% | 67% |
| LMArena Expert | 1521 | 1259 |
| Humanity's Last Exam | — | 5.7% |
| SimpleQA Verified | 50.6% | — |
| MMLU-Pro | — | 81% |
| Confabulations | — | 22.6% |
| Vectara Hallucination Rate | — | 8.2% |
| GPQA (HELM) | — | 65% |
Multimodal Kimi K3 leads
Kimi K3: 37.8 (#70), Llama 4 Maverick: 31.6 (#105)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| LMArena Vision | — | 1142 |
| GeoBench | — | 52% |
| Blueprint-Bench 2 | 29.5% | — |
| Furniture Assembly | 34.2% | — |
| SpatialViz-Bench | — | 31.8% |
Multilingual Kimi K3 leads
Kimi K3: 56.3 (#21), Llama 4 Maverick: 42.2 (#195)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| LMArena Non-English | 1466 | 1269 |
| LMArena Chinese | 1529 | 1277 |
| LMArena French | 1491 | 1259 |
| LMArena German | 1488 | 1291 |
| LMArena Japanese | 1487 | 1207 |
| LMArena Korean | 1458 | 1203 |
| LMArena Russian | 1482 | 1286 |
| LMArena Spanish | 1472 | 1293 |
Instruction Following Kimi K3 leads
Kimi K3: 77.7 (#14), Llama 4 Maverick: 71.7 (#146)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| LMArena Instruction Following | 1483 | 1267 |
| IFEval | — | 90.8% |
Long Context Kimi K3 leads
Kimi K3: 45.8 (#29), Llama 4 Maverick: 31.4 (#279)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| LMArena Longer Query | 1494 | 1280 |
| Fiction.LiveBench | — | 46.2% |
Writing & Preference Kimi K3 leads
Kimi K3: 76.6 (#4), Llama 4 Maverick: 38.8 (#252)
| Benchmark | Kimi K3 | Llama 4 Maverick |
|---|---|---|
| LMArena Text | 1476 | 1287 |
| LMArena Creative Writing | 1454 | 1267 |
| EQ-Bench Creative Writing | 2082 | 860 |
| LMArena Multi-Turn | 1488 | 1289 |
| Short-Story Creative Writing | — | 62% |
| WildBench | — | 80% |
| EQ-Bench 4 | 1339 | — |
Frequently asked questions
Is Kimi K3 better than Llama 4 Maverick?
Kimi K3 is the stronger model overall, scoring 59.5 to 30.9 on the Noometry Index. Llama 4 Maverick costs 20× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Which is cheaper, Kimi K3 or Llama 4 Maverick?
Llama 4 Maverick is cheaper. It lists at $0.19 per million input tokens and $0.65 per million output tokens; Kimi K3 lists at $3 and $15.
Is Kimi K3 or Llama 4 Maverick better for coding?
Kimi K3 scores higher on coding benchmarks: 61.0 versus 26.6 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 128K.
How many benchmarks do Kimi K3 and Llama 4 Maverick share?
32 benchmarks have published results for both models. Kimi K3 has 53 scored results on Noometry and Llama 4 Maverick has 54.