Both entries come from the live model catalog and use the same tier labels as their model pages. When a cached-input rate or a catalog entry is missing, this page says so instead of substituting zero or a hardcoded price.
Qwen4 Max vs Qwen3.8 Max
A catalog comparison of context, output ceilings and CNY per-million-token prices. Qwen4 Max has no results in the site's historical case data, so no same-case performance comparison is possible.
Long inputs, large outputs
Check the catalog limits and the input tier that applies to your request size. A large context or output ceiling increases capacity, not guaranteed task completion; validate quality and cost on your own workload.
Flagship Qwen3.8 option
Compare the current catalog rate and limits with the workload you need to run. Test accuracy and latency on your own inputs before moving production traffic.
Context, output and price by tier
Every value below is read from the same catalog entry that powers the model pages. Prices are in CNY per million tokens.
Tier labels preserve the catalog's input-size and time-band conditions. A missing cached-input price is shown as missing, not as zero.
| Model | Tier / condition | Context | Max output | Input / 1M | Cached input / 1M | Output / 1M |
|---|---|---|---|---|---|---|
| Qwen4 MaxReasoning · 1M context | Standard | 1M | 131K | ¥8.5 | ¥0.425 | ¥25.5 |
| Qwen3.8 MaxChat · 991K context | Standard | 991K | 131K | ¥6 | ¥0.75 | ¥18 |
The catalog rate applicable when a request starts is authoritative. Displayed prices already include the configured discount.
Tier labels preserve the catalog's input-size and time-band conditions. A missing cached-input price is shown as missing, not as zero.
No same-case results for Qwen4 Max
/compare/data contains no Qwen4 Max records.
Because Qwen4 Max has no result for the same cases, this page reports no cross-model score, pass rate, latency comparison or ranking, and does not reinterpret older samples as a Qwen4 Max result.
Questions about this comparison
Does this page rank Qwen4 Max against Qwen3.8 Max?
No. It compares catalog limits and prices only. Same-case measurements for Qwen4 Max are not available, so no performance ranking is made.
What happens if a price is missing?
The table marks the missing cached-input price or the missing catalog entry instead of showing zero, and no substitute rate is invented.
Which context and output limits are shown?
The values from each current catalog entry. Check both against your prompt and output budget; a limit describes capacity, not guaranteed output quality.
Which rate applies to a request?
The rate for the input-size and time-band tier named in the table, in CNY per million tokens. The platform rate when the request starts is authoritative.
Compare on your own workload
Both models share one API key when the live catalog lists them. Start with the workload you actually need to run.