Model comparison
Kimi K2.5 vs Llama 4 Maverick
Kimi K2.5 is the stronger model overall, scoring 48.1 to 30.9 on the Noometry Index. Llama 4 Maverick costs 3.0× less per token, which makes it the better buy when Kimi K2.5's lead doesn't matter for your workload.
Last verified . 37 shared benchmarks.
MetaSummary
- They share 37 benchmarks with published results for both. Kimi K2.5 scores higher in 10 categories and Llama 4 Maverick in 0 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Kimi K2.5 leads 65.1 to 38.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 92.2% for Kimi K2.5 and 20.6% for Llama 4 Maverick.
- Llama 4 Maverick is cheaper at $0.19 / $0.65 per million input/output tokens, against $0.45 / $2.25 for Kimi K2.5.
- Kimi K2.5 accepts more context: 262K tokens versus 128K.
Side by side
| Kimi K2.5 | Llama 4 Maverick | |
|---|---|---|
| Provider | Moonshot AI | Meta |
| Noometry Index | 48.1 | 30.9 |
| Released | 2026-01-27 | 2025-04-05 |
| Weights | Open | Open |
| Context window | 262K | 128K |
| Max output | 262K | 4K |
| Input $ / M tokens | $0.45 | $0.19 |
| Output $ / M tokens | $2.25 | $0.65 |
| Results tracked | 51 | 54 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.5 leads
Kimi K2.5: 48.8 (#53), Llama 4 Maverick: 26.6 (#324)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| SWE-bench Verified (bash only) | 70.8% | 21% |
| SciCode | 49% | 33.1% |
| WeirdML | 45.6% | 24.5% |
| LMArena Coding | 1474 | 1302 |
| ALE-Bench | 821.65 | 172.97 |
| SWE-bench Verified | 73.8% | — |
| Aider Polyglot | — | 15.6% |
| LMArena WebDev | 1437 | — |
| SWE-bench Multilingual | 67.3% | — |
| BigCodeBench Instruct | — | 49.7% |
| BigCodeBench Complete | — | 61.4% |
Agentic & Tool Use Kimi K2.5 leads
Kimi K2.5: 34.2 (#48), Llama 4 Maverick: 28.2 (#91)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| Terminal-Bench | 43.2% | — |
| Berkeley Function Calling Leaderboard | — | 37.3% |
| OSWorld | 63.3% | — |
| Vending-Bench 2 | 1,198 | — |
Reasoning Kimi K2.5 leads
Kimi K2.5: 31.2 (#80), Llama 4 Maverick: 10.1 (#342)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| ARC-AGI-2 | 11.8% | 0% |
| SimpleBench | 46.8% | 27.7% |
| Kagi LLM Benchmark | 78.5% | 55.9% |
| NYT Connections (extended) | 69.9% | 8% |
| ARC-AGI-1 | 65.3% | 4.4% |
| CritPt | 3.1% | 0% |
| EnigmaEval | 3.4% | 0.6% |
| LMArena Hard Prompts | 1453 | 1281 |
| Epoch Capabilities Index | 148.03 | 132.2 |
| Chess Puzzles | 12% | — |
| Thematic Generalization | 69.4% | — |
| DTBench | — | 61.9% |
| LMCA | — | 15.9% |
| ForecastBench | — | 57.5 |
Math Kimi K2.5 leads
Kimi K2.5: 51.8 (#53), Llama 4 Maverick: 26.0 (#262)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 92.2% | 20.6% |
| LMArena Math | 1470 | 1299 |
| FrontierMath (Feb 2025 set) | 27.9% | 0.7% |
| MathArena Final-Answer Competitions | 62.3% | — |
| Omni-MATH | — | 42.2% |
| MATH Level 5 | — | 73% |
| FrontierMath Tier 4 (v1) | 4.2% | — |
Knowledge Kimi K2.5 leads
Kimi K2.5: 53.6 (#56), Llama 4 Maverick: 33.4 (#204)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| GPQA Diamond | 87.6% | 67% |
| Humanity's Last Exam | 24.4% | 5.7% |
| Vectara Hallucination Rate | 14.2% | 8.2% |
| LMArena Expert | 1466 | 1259 |
| SimpleQA Verified | 34.3% | — |
| MMLU-Pro | — | 81% |
| Confabulations | — | 22.6% |
| GPQA (HELM) | — | 65% |
Multimodal Kimi K2.5 leads
Kimi K2.5: 41.1 (#39), Llama 4 Maverick: 31.6 (#105)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| LMArena Vision | 1269 | 1142 |
| GeoBench | — | 52% |
| LMArena Document | 1430 | — |
| SpatialViz-Bench | — | 31.8% |
Multilingual Kimi K2.5 leads
Kimi K2.5: 53.9 (#53), Llama 4 Maverick: 42.2 (#195)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| LMArena Non-English | 1433 | 1269 |
| LMArena Chinese | 1495 | 1277 |
| LMArena French | 1454 | 1259 |
| LMArena German | 1441 | 1291 |
| LMArena Japanese | 1421 | 1207 |
| LMArena Korean | 1410 | 1203 |
| LMArena Russian | 1435 | 1286 |
| LMArena Spanish | 1450 | 1293 |
Instruction Following Kimi K2.5 leads
Kimi K2.5: 75.3 (#64), Llama 4 Maverick: 71.7 (#146)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| LMArena Instruction Following | 1431 | 1267 |
| IFEval | — | 90.8% |
Long Context Kimi K2.5 leads
Kimi K2.5: 52.1 (#7), Llama 4 Maverick: 31.4 (#279)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| Fiction.LiveBench | 86.1% | 46.2% |
| LMArena Longer Query | 1445 | 1280 |
| CL-bench | 19.3% | — |
| CL-bench Life | 13.2% | — |
Writing & Preference Kimi K2.5 leads
Kimi K2.5: 65.1 (#53), Llama 4 Maverick: 38.8 (#252)
| Benchmark | Kimi K2.5 | Llama 4 Maverick |
|---|---|---|
| LMArena Text | 1445 | 1287 |
| LMArena Creative Writing | 1423 | 1267 |
| EQ-Bench Creative Writing | 1579 | 860 |
| LMArena Multi-Turn | 1444 | 1289 |
| Short-Story Creative Writing | — | 62% |
| WildBench | — | 80% |
Frequently asked questions
Is Kimi K2.5 better than Llama 4 Maverick?
Kimi K2.5 is the stronger model overall, scoring 48.1 to 30.9 on the Noometry Index. Llama 4 Maverick costs 3.0× less per token, which makes it the better buy when Kimi K2.5's lead doesn't matter for your workload.
Which is cheaper, Kimi K2.5 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 K2.5 lists at $0.45 and $2.25.
Is Kimi K2.5 or Llama 4 Maverick better for coding?
Kimi K2.5 scores higher on coding benchmarks: 48.8 versus 26.6 in the Noometry coding category.
Which has the bigger context window?
Kimi K2.5 does, with 262K tokens against 128K.
How many benchmarks do Kimi K2.5 and Llama 4 Maverick share?
37 benchmarks have published results for both models. Kimi K2.5 has 51 scored results on Noometry and Llama 4 Maverick has 54.