# Qwen3.8 Max vs Qwen4 Max

As of November 17, 2026, Qwen3.8 Max costs 33% more per input token than Qwen4 Max ($2.00 vs $1.50 per 1M), and both have a 992K-token context window.

- Page: https://models.litellm.ai/compare/qwen3.8-max-vs-qwen4-max
- Qwen3.8 Max: https://models.litellm.ai/models/qwen3.8-max
- Qwen4 Max: https://models.litellm.ai/models/qwen4-max

## Side by side

| | Qwen3.8 Max | Qwen4 Max |
|---|---|---|
| Input, per 1M tokens | $2.00 | $1.50 |
| Output, per 1M tokens | $6.00 | $4.50 |
| Cache read, per 1M tokens | $0.25 | $0.075 |
| Cache write, per 1M tokens | – | $1.875 |
| Context window | 991,808 tokens | 991,808 tokens |
| Max output | 131,072 tokens | 131,072 tokens |
| Image input | Yes | Yes |
| Tool calling | Yes | Yes |
| Structured output (JSON) | Yes | Yes |
| Prompt caching | Yes | Yes |
| Reasoning | Yes | Yes |
| Providers on LiteLLM | 12 | 1 |
| Price shown is from | Alibaba Cloud Model Studio | Alibaba Cloud Model Studio |
| Added to LiteLLM | August 7, 2026 | November 16, 2026 |

Headline prices in USD per 1M tokens, from LiteLLM's model price file.

## Cheapest provider for each

- Qwen3.8 Max: DeepInfra at $1.65 input per 1M tokens (`deepinfra/Qwen/Qwen3.8-Max`)
- Qwen4 Max: Alibaba Cloud Model Studio at $1.50 input per 1M tokens (`dashscope/qwen4-max`)

## When to pick which

- Qwen3.8 Max: Available from 12 providers through LiteLLM, vs 1.
- Qwen4 Max: Costs 25% less per input token: $1.50 vs $2.00 per 1M.

## FAQ

### Which is cheaper, Qwen3.8 Max or Qwen4 Max?

Qwen4 Max is cheaper at 3 input tokens per output token: $2.25 per 1M tokens blended ($1.50 input, $4.50 output), versus $3.00 ($2.00 and $6.00) for Qwen3.8 Max, as of November 17, 2026.

### Which has the larger context window, Qwen3.8 Max or Qwen4 Max?

Both have a 991,808-token context window.

### Can I switch from Qwen3.8 Max to Qwen4 Max without changing my code?

Yes. Through LiteLLM both use the OpenAI format, so switching means changing the model name from "dashscope/qwen3.8-max" to "dashscope/qwen4-max", or pointing a LiteLLM gateway model alias at the other model. You can also use one as a fallback for the other.
