Qwen Pricing: Which Model Is Worth Paying For?
OpenRouter lists 54 Qwen models. For 1M input plus 1M output tokens they range from $0.16 (Qwen3.7 Flash) to $16 (Qwen3.8 Max Prime), a 100x spread. Most teams overpay by defaulting to the flagship for work a cheaper model passes.
Plans and pricing
1 free Qwen model
$0 on OpenRouter free variantsFree variants such as qwen/qwen3.8-27b:free. Free tiers are rate-limited and can change without notice.
Best for
Prototyping and benchmarking before you commit spend.
Not for
Production traffic that needs predictable throughput.
Qwen3.7 Flash
$0.03 input / $0.13 output per 1M tokens1M context on OpenRouter (qwen/qwen3.7-flash). 1M input plus 1M output tokens: $0.16.
Best for
High-volume, low-stakes steps: routing, tagging, extraction and first drafts.
Not for
Tasks where a wrong answer costs more than the tokens saved.
Qwen3 Next 80B A3B Instruct
$0.10 input / $1.10 output per 1M tokens262K context on OpenRouter (qwen/qwen3-next-80b-a3b-instruct). 1M input plus 1M output tokens: $1.20.
Best for
A middle tier to test before paying flagship prices.
Not for
Tasks where a wrong answer costs more than the tokens saved.
Qwen3.8 2.4T A95B
$2 input / $6 output per 1M tokens1M context on OpenRouter (qwen/qwen3.8-2.4t-a95b). 1M input plus 1M output tokens: $8.
Best for
A middle tier to test before paying flagship prices.
Not for
Tasks where a wrong answer costs more than the tokens saved.
Qwen3.8 Max Prime
$4 input / $12 output per 1M tokens1M context on OpenRouter (qwen/qwen3.8-max-prime). 1M input plus 1M output tokens: $16.
Best for
The hardest Qwen tasks, once cheaper Qwen models have failed your acceptance tests.
Not for
Bulk work a smaller model already passes; the price gap compounds at volume.
Prices are OpenRouter's listed per-token prices for Alibaba Cloud (Qwen) models, read from the OpenRouter models API and checked 2026-10-04. Buying direct from Alibaba Cloud (Qwen) or another host can be priced differently; prices exclude tax and change often.
Who should pay for Qwen?
Qwen suits teams running high-volume chat, extraction or agent steps who want a low-cost open-weight family with very long context. Start with Qwen3.7 Flash on a sample of real tasks, measure the acceptance rate, and step up a tier only where it fails. Price the whole job, not the token: retries and human review often cost more than the model.
Calculate your exact cost →Who is overpaying for Qwen?
- Teams sending every request to Qwen3.8 Max Prime when Qwen3.7 Flash passes most of them
- Anyone paying for output tokens they never read: cap max tokens and ask for concise answers
- Buyers comparing token prices without counting retries, failed responses and review time
Official pricing sources
Frequently asked questions
What is the cheapest Qwen model?
On OpenRouter, Qwen3.7 Flash at $0.03 input and $0.13 output per 1M tokens (checked 2026-10-04); 1 free variant is also listed, rate-limited.
How much does the top Qwen model cost?
Qwen3.8 Max Prime is listed at $4 input and $12 output per 1M tokens, so 1M of each costs $16.
Is Qwen cheaper than GPT or Claude?
Usually per token, yes. Compare on your own prompts with the AI Cost Calculator: the cheaper model only saves money if it passes the same acceptance tests.