Model comparison
Kimi K2 (Jul 2025) vs Qwen3.8 27B
Qwen3.8 27B is the stronger model overall, scoring 46.0 to 41.2 on the Noometry Index.
Last verified . 19 shared benchmarks.
Alibaba (Qwen)Summary
- They share 19 benchmarks with published results for both. Kimi K2 (Jul 2025) scores higher in 1 category and Qwen3.8 27B in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Qwen3.8 27B leads 41.0 to 23.3.
- Kimi K2 (Jul 2025) is cheaper at $0.57 / $2.30 per million input/output tokens, against $0.99 / $1.49 for Qwen3.8 27B.
Side by side
| Kimi K2 (Jul 2025) | Qwen3.8 27B | |
|---|---|---|
| Provider | Moonshot AI | Alibaba (Qwen) |
| Noometry Index | 41.2 | 46.0 |
| Released | 2025-07-12 | 2026-08-14 |
| Weights | Open | Open |
| Context window | 262K | 262K |
| Max output | 262K | 33K |
| Input $ / M tokens | $0.57 | $0.99 |
| Output $ / M tokens | $2.30 | $1.49 |
| Results tracked | 42 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 27B leads
Kimi K2 (Jul 2025): 42.4 (#102), Qwen3.8 27B: 50.5 (#44)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| LMArena Coding | 1399 | 1482 |
| SWE-bench Verified (bash only) | 63.4% | — |
| Aider Polyglot | 59.1% | — |
| LMArena WebDev | — | 1593 |
| SciCode | — | 46.6% |
| GSO | 4.9% | — |
| WeirdML | 42.8% | — |
| ALE-Bench | 597.5 | — |
Agentic & Tool Use Too close to call
Kimi K2 (Jul 2025): 32.4 (#64), Qwen3.8 27B: 32.9 (#57)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| Terminal-Bench | 35.7% | — |
| APEX-Agents | — | 47.5% |
| Berkeley Function Calling Leaderboard | 59.1% | — |
| METR Time Horizons | 59.2% | — |
Reasoning Qwen3.8 27B leads
Kimi K2 (Jul 2025): 23.3 (#179), Qwen3.8 27B: 41.0 (#54)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| LMArena Hard Prompts | 1384 | 1460 |
| Epoch Capabilities Index | 146.01 | 149.38 |
| ARC-AGI-2 | — | 42.4% |
| SimpleBench | 26.3% | — |
| Kagi LLM Benchmark | 64.4% | — |
| NYT Connections (extended) | — | 54.5% |
| ARC-AGI-1 | — | 87.5% |
| CritPt | — | 5.4% |
| DTBench | — | 88% |
| LMCA | — | 41.4% |
| Surface Evolver Bench | — | 45% |
| ForecastBench | 60.2 | — |
Math Kimi K2 (Jul 2025) leads
Kimi K2 (Jul 2025): 42.7 (#83), Qwen3.8 27B: 37.1 (#161)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| LMArena Math | 1397 | 1456 |
| ProofBench | — | 16% |
| Omni-MATH | 65.4% | — |
| FrontierMath (Feb 2025 set) | 21.4% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge Qwen3.8 27B leads
Kimi K2 (Jul 2025): 37.3 (#157), Qwen3.8 27B: 41.6 (#109)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| LMArena Expert | 1365 | 1482 |
| MMLU-Pro | 81.9% | — |
| Confabulations | 20.4% | — |
| Vectara Hallucination Rate | 17.9% | — |
| GPQA (HELM) | 65.3% | — |
Multimodal Not comparable
Kimi K2 (Jul 2025): —, Qwen3.8 27B: 41.3 (#37)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| LMArena Vision | — | 1271 |
Multilingual Qwen3.8 27B leads
Kimi K2 (Jul 2025): 49.6 (#130), Qwen3.8 27B: 53.7 (#60)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| LMArena Non-English | 1372 | 1430 |
| LMArena Chinese | 1415 | 1504 |
| LMArena French | 1379 | 1465 |
| LMArena German | 1387 | 1438 |
| LMArena Japanese | 1349 | 1384 |
| LMArena Korean | 1325 | 1393 |
| LMArena Russian | 1385 | 1415 |
| LMArena Spanish | 1386 | 1448 |
Instruction Following Qwen3.8 27B leads
Kimi K2 (Jul 2025): 71.1 (#156), Qwen3.8 27B: 75.8 (#53)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| LMArena Instruction Following | 1348 | 1439 |
| IFEval | 85% | — |
Long Context Qwen3.8 27B leads
Kimi K2 (Jul 2025): 41.2 (#145), Qwen3.8 27B: 44.3 (#70)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| LMArena Longer Query | 1353 | 1450 |
| Fiction.LiveBench | 66.7% | — |
| CL-bench | 17.6% | — |
Writing & Preference Qwen3.8 27B leads
Kimi K2 (Jul 2025): 62.3 (#78), Qwen3.8 27B: 65.8 (#43)
| Benchmark | Kimi K2 (Jul 2025) | Qwen3.8 27B |
|---|---|---|
| LMArena Text | 1380 | 1441 |
| LMArena Creative Writing | 1350 | 1384 |
| EQ-Bench Creative Writing | 1666 | 1671 |
| LMArena Multi-Turn | 1371 | 1441 |
| Short-Story Creative Writing | 85.6% | — |
| WildBench | 86.2% | — |
Frequently asked questions
Is Kimi K2 (Jul 2025) better than Qwen3.8 27B?
Qwen3.8 27B is the stronger model overall, scoring 46.0 to 41.2 on the Noometry Index.
Which is cheaper, Kimi K2 (Jul 2025) or Qwen3.8 27B?
Kimi K2 (Jul 2025) is cheaper. It lists at $0.57 per million input tokens and $2.30 per million output tokens; Qwen3.8 27B lists at $0.99 and $1.49.
Is Kimi K2 (Jul 2025) or Qwen3.8 27B better for coding?
Qwen3.8 27B scores higher on coding benchmarks: 50.5 versus 42.4 in the Noometry coding category.
Which has the bigger context window?
Both accept 262K tokens.
How many benchmarks do Kimi K2 (Jul 2025) and Qwen3.8 27B share?
19 benchmarks have published results for both models. Kimi K2 (Jul 2025) has 42 scored results on Noometry and Qwen3.8 27B has 31.