Qwen
Qwen is Alibaba Cloud’s family of large language models and AI platform, spanning open models and multimodal services.
Pros
- Qwen3.8-Max is a 2.4-trillion-parameter flagship released on 2 August 2026 with a 1,000,000-token context window (991,808 in, 131,072 out) at $2.00/M input, $0.25/M cached input and $6.00/M output, roughly a third of what comparable Western flagships charge, and Alibaba open-sources aggressively, publishing Max-class weights for the first time with Qwen3.8-Max alongside open releases of Qwen3.8-Flash-Next, Qwen-Image-3.0 and Qwen-AgentWorld. The cheaper tiers are unusually strong: qwen3.8-flash runs $0.113/M input and $0.382/M output across most regions and still carries the full 1M-token window, while qwen3.7-plus starts at $0.40/M input and $1.60/M output. The consumer side, Qwen Studio at qwen.ai, is free with no subscription and covers thinking, web search, Deep Research, Web Dev, Artifacts, image editing and generation and video generation on web, iOS, Android, macOS and Windows
- the API is OpenAI-compatible so migration is a base-URL change, and new Model Studio accounts get 1M free tokens per model for 90 days.
Cons
- Cost modelling is genuinely awkward: rates differ by region, with qwen3.8-flash at $0.15/M input and $0.47/M output on the Singapore endpoint but $0.113/$0.382 in Beijing, Hong Kong, Frankfurt, Tokyo and US Virginia, and the Plus and Flash lines bill in context bands, so qwen3.7-plus jumps from $0.40/$1.60 below 256K to $1.20/$4.80 above it and a long-context workload can cost triple the headline rate. The free allowance is 1M tokens per model for 90 days and cannot be pooled or transferred between models, and the free developer tiers around Qwen Code have been cut and reinstated at short notice. There is no consumer subscription, so Qwen Studio's free limits are the only limits and they are published nowhere. Buying the API means an Alibaba Cloud Model Studio account with region selection and identity verification rather than a self-serve key, and version churn is fast, with Qwen3.5, 3.6, 3.7 and 3.8 all shipping inside a year, which makes pinning model IDs mandatory.
Cost
Qwen Studio at qwen.ai is free with no paid consumer tier. API billing runs through Alibaba Cloud Model Studio, pay-as-you-go per token and priced by region: qwen3.8-max $2.00/M input, $0.25/M cached, $6.00/M output on the Singapore endpoint ($1.65/$0.206/$4.951 in Beijing) at a 1M-token context; qwen3.8-flash $0.15/M input and $0.47/M output in Singapore, $0.113/$0.382 with $0.014 cached elsewhere; qwen3.7-plus tiered at $0.40/$1.60 up to 256K and $1.20/$4.80 from 256K to 1M; qwen3.7-flash from $0.030/$0.130 at short context; qwen3.5-flash $0.10/$0.40; qwen3-max tiered from $1.20/$6.00 to $3.00/$15.00. New users get 1M free tokens per model, valid 90 days from activation or model release. Open-weight releases such as Qwen3.8-Flash-Next cost nothing to run on your own hardware.
Verdict
Best for developers and platform teams who want frontier-adjacent quality at roughly a third of Western flagship prices, or an open-weight family they can self-host and fine-tune; weigh it carefully if you need flat predictable billing, a paid consumer plan, or you cannot put workloads on Alibaba Cloud.
Visit Qwen