Kimi K2 Thinking is slightly cheaper per million tokens (blended). Qwen3.8 27B has the larger context window (1,000,000 vs 262,144 tokens, 3.8×).
Kimi K2 Thinking vs Qwen3.8 27B
Live catalog fields. Best-in-row is highlighted.
Kimi K2 Thinkingopen
Moonshotai · moonshotai__kimi-k2-thinking
Qwen3.8 27Bopen
Qwen · qwen__qwen3.8-27b
Identity
| Field | moonshotai__kimi-k2-thinking | qwen__qwen3.8-27b |
|---|---|---|
| Provider | Moonshotai | Qwen |
| Slug | moonshotai__kimi-k2-thinking | qwen__qwen3.8-27b |
| Status | live | live |
| Open weights | Yes | Yes |
| License | — | — |
| HuggingFace | moonshotai/Kimi-K2-Thinking | Qwen/Qwen3.8-27B |
Quality
| Field | moonshotai__kimi-k2-thinking | qwen__qwen3.8-27b |
|---|---|---|
| Quality index (0–100) | — | 89.0 |
Cost
| Field | moonshotai__kimi-k2-thinking | qwen__qwen3.8-27b |
|---|---|---|
| Input $ / 1M tokens | $0.60 | $0.42 |
| Cached input $ / 1M | $0.15 | $0.09 |
| Output $ / 1M tokens | $2.50 | $3.00 |
Context
| Field | moonshotai__kimi-k2-thinking | qwen__qwen3.8-27b |
|---|---|---|
| Max context tokens | 262,144 | 1,000,000 |
| Max output tokens | 100,352 | 131,072 |
Modalities
| Field | moonshotai__kimi-k2-thinking | qwen__qwen3.8-27b |
|---|---|---|
| Input | text | text, image, video |
| Output | text | text |