Model comparison
Kimi K3 vs Qwen3.8 27B
Kimi K3 is the stronger model overall, scoring 59.5 to 46.0 on the Noometry Index. Qwen3.8 27B costs 5.4× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Last verified . 30 shared benchmarks.
Alibaba (Qwen)Summary
- They share 30 benchmarks with published results for both. Kimi K3 scores higher in 9 categories and Qwen3.8 27B in 1 category; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K3 leads 74.2 to 37.1.
- The biggest single-benchmark swing is ProofBench: 87% for Kimi K3 and 16% for Qwen3.8 27B.
- Qwen3.8 27B is cheaper at $0.99 / $1.49 per million input/output tokens, against $3 / $15 for Kimi K3.
- Kimi K3 accepts more context: 1.05M tokens versus 262K.
Side by side
| Kimi K3 | Qwen3.8 27B | |
|---|---|---|
| Provider | Moonshot AI | Alibaba (Qwen) |
| Noometry Index | 59.5 | 46.0 |
| Released | 2026-07-16 | 2026-08-14 |
| Weights | Open | Open |
| Context window | 1.05M | 262K |
| Max output | 1.05M | 33K |
| Input $ / M tokens | $3 | $0.99 |
| Output $ / M tokens | $15 | $1.49 |
| Results tracked | 53 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K3 leads
Kimi K3: 61.0 (#10), Qwen3.8 27B: 50.5 (#44)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| LMArena WebDev | 1654 | 1593 |
| SciCode | 59.5% | 46.6% |
| LMArena Coding | 1508 | 1482 |
| DeepSWE | 68.5% | — |
| FrontierCode | 44.2% | — |
| FrontierSWE | 25.9% | — |
| WeirdML | 82.6% | — |
| ALE-Bench | 1,524 | — |
Agentic & Tool Use Kimi K3 leads
Kimi K3: 41.8 (#20), Qwen3.8 27B: 32.9 (#57)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| APEX-Agents | 50.6% | 47.5% |
| 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), Qwen3.8 27B: 41.0 (#54)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| ARC-AGI-2 | 60.4% | 42.4% |
| NYT Connections (extended) | 93.6% | 54.5% |
| ARC-AGI-1 | 94.5% | 87.5% |
| CritPt | 23.4% | 5.4% |
| LMArena Hard Prompts | 1496 | 1460 |
| DTBench | 91.2% | 88% |
| LMCA | 52.7% | 41.4% |
| Surface Evolver Bench | 95% | 45% |
| Epoch Capabilities Index | 157.45 | 149.38 |
| SimpleBench | 60.7% | — |
| Chess Puzzles | 39% | — |
| Mystery Game Puzzles | 26% | — |
| ForecastBench | 61.1 | — |
Math Kimi K3 leads
Kimi K3: 74.2 (#16), Qwen3.8 27B: 37.1 (#161)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| ProofBench | 87% | 16% |
| LMArena Math | 1491 | 1456 |
| FrontierMath (Tiers 1-3) | 72.2% | — |
| FrontierMath Tier 4 | 39% | — |
| MathArena Final-Answer Competitions | 87.8% | — |
| OTIS Mock AIME 2024-2025 | 97.2% | — |
Knowledge Kimi K3 leads
Kimi K3: 63.2 (#21), Qwen3.8 27B: 41.6 (#109)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| LMArena Expert | 1521 | 1482 |
| GPQA Diamond | 93.1% | — |
| SimpleQA Verified | 50.6% | — |
Multimodal Qwen3.8 27B leads
Kimi K3: 37.8 (#70), Qwen3.8 27B: 41.3 (#37)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| LMArena Vision | — | 1271 |
| Blueprint-Bench 2 | 29.5% | — |
| Furniture Assembly | 34.2% | — |
Multilingual Kimi K3 leads
Kimi K3: 56.3 (#21), Qwen3.8 27B: 53.7 (#60)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| LMArena Non-English | 1466 | 1430 |
| LMArena Chinese | 1529 | 1504 |
| LMArena French | 1491 | 1465 |
| LMArena German | 1488 | 1438 |
| LMArena Japanese | 1487 | 1384 |
| LMArena Korean | 1458 | 1393 |
| LMArena Russian | 1482 | 1415 |
| LMArena Spanish | 1472 | 1448 |
Instruction Following Kimi K3 leads
Kimi K3: 77.7 (#14), Qwen3.8 27B: 75.8 (#53)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| LMArena Instruction Following | 1483 | 1439 |
Long Context Kimi K3 leads
Kimi K3: 45.8 (#29), Qwen3.8 27B: 44.3 (#70)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| LMArena Longer Query | 1494 | 1450 |
Writing & Preference Kimi K3 leads
Kimi K3: 76.6 (#4), Qwen3.8 27B: 65.8 (#43)
| Benchmark | Kimi K3 | Qwen3.8 27B |
|---|---|---|
| LMArena Text | 1476 | 1441 |
| LMArena Creative Writing | 1454 | 1384 |
| EQ-Bench Creative Writing | 2082 | 1671 |
| LMArena Multi-Turn | 1488 | 1441 |
| EQ-Bench 4 | 1339 | — |
Frequently asked questions
Is Kimi K3 better than Qwen3.8 27B?
Kimi K3 is the stronger model overall, scoring 59.5 to 46.0 on the Noometry Index. Qwen3.8 27B costs 5.4× 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 Qwen3.8 27B?
Qwen3.8 27B is cheaper. It lists at $0.99 per million input tokens and $1.49 per million output tokens; Kimi K3 lists at $3 and $15.
Is Kimi K3 or Qwen3.8 27B better for coding?
Kimi K3 scores higher on coding benchmarks: 61.0 versus 50.5 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 262K.
How many benchmarks do Kimi K3 and Qwen3.8 27B share?
30 benchmarks have published results for both models. Kimi K3 has 53 scored results on Noometry and Qwen3.8 27B has 31.