GLM-5.3 vs Kimi K2.6
GLM-5.3 (Z.ai) costs $1.40 per million input tokens and $4.40 per million output tokens; Kimi K2.6 (Moonshot AI) costs $0.95 and $4.00. Kimi K2.6 is 1.3× cheaper for a typical 3:1 input-to-output mix. GLM-5.3 has the larger context window (1M vs 262K tokens).
| GLM-5.3 | Kimi K2.6 | |
|---|---|---|
| Provider | Z.ai | Moonshot AI |
| Input / 1M tokens | $1.40 | $0.95 |
| Output / 1M tokens | $4.40 | $4.00 |
| Cached input / 1M | $0.26 | $0.16 |
| Blended (3:1) | $2.15 | $1.71 |
| Context window | 1M | 262K |
| Max output | 128K | 262K |
| Vision | — | Yes |
| Tool use | Yes | Yes |
| Reasoning | Yes | Yes |
| Listed since | Aug 30, 2026 | Apr 25, 2026 |
Cost per 1,000 requests
| Workload | GLM-5.3 | Kimi K2.6 |
|---|---|---|
| Chatbot reply2,000 input + 500 output tokens | $5.00 | $3.90 |
| RAG / search answer8,000 input + 700 output tokens | $14.28 | $10.40 |
| Long-document summary100,000 input + 2,000 output tokens | $148.80 | $103.00 |
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
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.