DeepSeek, Qwen and GLM API pricing compared (per million tokens)
Published · Updated
Token prices differ by a large factor between the models you can reach through one key. This page compares them using live prices from the ModelSite catalog, so the table is never out of date.
Price table
| Model | Input | Output | Cache read | Context |
|---|---|---|---|---|
| DeepSeek V4.1 Flash 1017 | $0.128 | $0.51 | $0.00255 | 1M |
| Qwen3.7-Max | $1.518 | $4.554 | $0.304 | 1M |
| Qwen3.8-Max | $1.697 | $5.089 | $0.213 | 1M |
| GLM-5.3-Flash | $0.072 | $0.25 | $0.021 | 1M |
| GLM-5.3 | $0.715 | $2.50 | $0.179 | 1M |
| MiniMax-M3 | $0.266 | $1.063 | $0.054 | 1M |
How big is the gap inside each family?
Qwen3.8-Max costs 1.1× as much per output token as Qwen3.7-Max.
GLM-5.3 costs 10.0× as much per output token as GLM-5.3-Flash.
Reading the table
- Output tokens are typically priced higher than input tokens, so long answers cost more than long prompts.
- Models that support prompt caching bill cached input at a lower rate; the cache column shows it when available.
- Some models have off-peak time windows with lower prices. The model page lists the windows.
Choosing by workload
For high-volume work such as classification or extraction, start from the cheapest row and measure quality on your own data. Move up a tier only where it measurably helps. Because every model sits behind the same endpoint, trying the next tier is a one-line change.
Estimate your bill
Use the cost calculator to turn these per-token prices into a monthly estimate, or open any model in the catalog for its full price breakdown.