Head to head
Qwen3.6 27B vs Grok 4.20 Non-Reasoning
On a blended workload of one million input tokens and three million output tokens, Grok 4.20 Non-Reasoning is 1.1× cheaper than Qwen3.6 27B.
| Qwen3.6 27B | Grok 4.20 Non-Reasoning | |
|---|---|---|
| Provider | Groq | xAI |
| Input / MTok | $0.6 | $1.25 |
| Output / MTok | $3 | $2.5 |
| Cache read / MTok | — | $0.2 |
| Context window | 131,072 | 1,000,000 |
| Blended 1M in + 3M out | $9.60 | $8.75 |
| Status | preview | ga |