Alibaba Cloud DashScope (Qwen) review

DashScope bills per token, with input from CNY 0.15 and output from CNY 0.50. Both figures are quoted in CNY, so a team on a dollar budget has to convert before comparing with providers that price in USD. Output is the costlier side of the pair, which matters for long generations.

Yes. 1,000,000 tokens free per model, valid for 90 days are granted per model, so a prototype can run a meaningful amount of traffic before the first bill. The allowance is time limited, so it rewards a quick first test rather than a long evaluation. Anything beyond it is charged at the standard token rates.

The largest models reach 1,000,000 tokens, which puts DashScope level with the 1,000,000 site median and well clear of the 256,000 lower quartile. That is enough for long documents in one call, and close to Azure AI Foundry's 1,050,000.

The API is OpenAI-compatible, so the same client libraries and request shape carry over. Teams generally swap the base URL and credentials rather than rewrite their code. Pair that with 1,000,000 tokens free per model, valid for 90 days and a model swap can be tested against real prompts before any spend is committed.

Salad lists input from $0.06 in dollars, while Together AI quotes $0 to $3. DashScope's CNY 0.15 is in CNY, so the headline numbers are not directly comparable without conversion. Teams that settle a currency first will find the comparison much easier.

Developers who want a Qwen model reachable through an OpenAI-compatible endpoint, and who need a long context window for document work. 1,000,000 tokens covers large files in one request. 1,000,000 tokens free per model, valid for 90 days also makes it a low cost place to run those first prompts.

On input price it sits 16 of 71 brands, and on the largest context window 21 of 72. The overall score is 6.8, split into cost 8 and offer 5.6. Cost carries the stronger part of that split.