Paid links may earn Compute Screener a referral fee. They never affect prices, weights or the order.
About Alibaba
The Alibaba token price index is $0.44 per 1M tokens today, down 21.3% since yesterday, down 18.5% over the past week and up 10.9% over the past month. Since Apr 11, 2026 it has ranged from $0.06 to $3.00. It is the usage-weighted average of the posted prices of Alibaba's 3 models, blending 3 input tokens with 1 output token; Qwen3.8 Flash has the largest share of usage, at 42.2%. It is calculated once a day, for each UTC day, shortly after the day ends.
Alibaba Cloud develops the Qwen models, first released in April 2023, and sells them through Alibaba Cloud Model Studio in Singapore, Beijing, Hong Kong, Frankfurt, Virginia and Tokyo, each with its own price list. Many Qwen models are open weights; there are more than 200,000 variants of them on Hugging Face. Among paid models, its list prices span the widest ratio of the Chinese companies on this site: Qwen3.7-Max's input price, $2.50 per million tokens, is about 80 times Qwen3.7-Flash's $0.03.
Models and list prices
| Model | Input | Output | 3:1 blend |
|---|
| Qwen3.7-Max | $2.50 | $7.50 | $3.75 |
| Qwen3.8-Max | $2.00 | $6.00 | $3.00 |
| Qwen3.8-27B, open weights | $0.50 | $3.00 | $1.125 |
| Qwen3.7-Plus, prompts up to 256K tokens | $0.40 | $1.60 | $0.70 |
| Qwen3.8-Flash | $0.15 | $0.47 | $0.23 |
| Qwen3.7-Flash, prompts up to 32K tokens | $0.03 | $0.13 | $0.055 |
USD per 1M tokens, international list for Singapore, on October 24, 2026, from Alibaba Cloud's Model Studio pricing page, last updated October 22, 2026. Many models are priced by request size, with every token in a longer request billed at the higher tier's rate. Qwen3.7-Plus carries a "Limited-time 20% off", batch calls cost half, cache hits cost 10% to 20% of the input price, and the Hong Kong region has night-time discounts on some models.
What this index measures
Each day's value averages the prices Alibaba posts for its own Qwen endpoints on OpenRouter, weighted by each model's tokens in OpenRouter's daily rankings and blended at 3 input tokens to 1 output token. Those prices can differ from Model Studio's list: on October 23, 2026 the index used $0.425/$2.55 for Qwen3.8-27B, 15% below the list. Open-weight Qwen models sold by other hosts count only while Alibaba serves them too.
Alibaba's index can move several-fold from one day to the next, because which Qwen models are among OpenRouter's 50 most-used changes from day to day. On October 8, 2026 only Qwen3.8-Max was, and the index was $3.00; on October 9 only the two Flash models were, and it was $0.14. These moves come from which models are counted, not from Alibaba changing prices.
Other hosts sell Qwen's open weights too. The open-weight version of Qwen3.8-Max requires model-as-a-service businesses with more than $50 million in revenue over 12 months to get a separate license, and all five hosts selling it on OpenRouter charged Alibaba's $2/$6 on October 24, 2026.
What moves Alibaba's index
- Which models make the top 50. In October 2026, to the 23rd, Qwen3.8-Flash carried 46% of the tokens the index counted for Alibaba, Qwen3.8-27B 35%, Qwen3.7-Flash 14% and Qwen3.8-Max 4%. A day when Max is counted and the Flash models are not lifts the index far above its usual level.
- Promotions and request size. Time-limited discounts change the posted price. Long requests can cost several times more per token than short ones, which the index, built from base rates, does not reflect.
- New flagships. On Model Studio's list, Qwen3.8-Max, at $2/$6, is cheaper than Qwen3.7-Max at $2.50/$7.50. The index, though, prices Qwen3.7-Max at the $1.475/$4.425 that Alibaba's endpoint lists on OpenRouter, so days that count Qwen3.8-Max, at a $3.00 blend, sit above days that counted Qwen3.7-Max, at $2.21.
Price history
- December 31, 2024: CNBC reported that Alibaba's cloud unit cut prices on AI models "by up to 85%".
- April 2025: Qwen3 released as open weights under the Apache 2.0 license.
- April 2026: Qwen3.6-Plus at $0.50/$3.
- May 2026: Qwen3.7-Max released; $2.50/$7.50 on today's list.
- August 2026: Qwen3.8-Max released, followed by its open weights on August 12; $2/$6 on today's list.
- August 26, 2026: Qwen3.8-Flash released; $0.15/$0.47 on today's list.
This index's history for Alibaba starts on April 11, 2026, and 59 days since then are empty under the index's rules, the last on August 16, 2026. On 56 of them, open-weight Qwen models that Alibaba's own API may or may not have served then carried more than 20% of its tokens; on the other 3, no Qwen model the index prices was among OpenRouter's 50 most-used.
US government actions
In June 2026 the US Department of Defense added Alibaba to its list of companies it says support China's military; listed companies are to be barred from consideration for US defense contracts. Alibaba said there was "no basis" for its inclusion, according to Al Jazeera.
FAQ
How much does the Qwen API cost per million tokens? From $0.03 input and $0.13 output for Qwen3.7-Flash on short prompts to $2.50 and $7.50 for Qwen3.7-Max; the newer Qwen3.8-Max costs $2 and $6, as in the table above.
Why does Alibaba's index jump between days? Which Qwen models are among OpenRouter's 50 most-used changes from day to day, and Qwen3.8-Max costs about 13 times as much as Qwen3.8-Flash on a 3:1 blend.
Why does the price depend on prompt length? For many Qwen models, the whole request is billed at the rate of its size tier, so long prompts cost more per token.
How often is it updated? Once a day: each UTC day's value is calculated shortly after that day ends, from the day's OpenRouter volumes and the prices posted then. The site checks for new data every 2 to 3 hours.
Sources
Text reviewed Oct 24, 2026; the figures in it are as of that date. The figures at the top of this page update once a day. Not investment advice.
