Kimi K3 vs Qwen3.8 Max
Kimi K3 (Moonshot AI) costs $3.00 per million input tokens and $15.00 per million output tokens; Qwen3.8 Max (Alibaba Cloud) costs $2.00 and $6.00. Qwen3.8 Max is 2.0× cheaper for a typical 3:1 input-to-output mix. Kimi K3 has the larger context window (1.05M vs 992K tokens).
| Kimi K3 | Qwen3.8 Max | |
|---|---|---|
| Provider | Moonshot AI | Alibaba Cloud |
| Input / 1M tokens | $3.00 | $2.00 |
| Output / 1M tokens | $15.00 | $6.00 |
| Cached input / 1M | $0.30 | $0.25 |
| Blended (3:1) | $6.00 | $3.00 |
| Context window | 1.05M | 992K |
| Max output | 1.05M | 131K |
| Vision | Yes | Yes |
| Tool use | Yes | Yes |
| Reasoning | Yes | Yes |
| Listed since | Aug 23, 2026 | Aug 16, 2026 |
Cost per 1,000 requests
| Workload | Kimi K3 | Qwen3.8 Max |
|---|---|---|
| Chatbot reply2,000 input + 500 output tokens | $13.50 | $7.00 |
| RAG / search answer8,000 input + 700 output tokens | $34.50 | $20.20 |
| Long-document summary100,000 input + 2,000 output tokens | $330.00 | $212.00 |
Blended price history
Blended = (3 × input + output) ÷ 4, per million tokens. Each line starts when the model first appeared in the price list.
The companies
Related comparisons
List prices for each provider's own API (USD), taken daily from LiteLLM's public model price list. Batch, caching and committed-use discounts aren't reflected. Price alone says nothing about output quality; check benchmarks and your own evaluations before switching.