LLM API pricing / Alibaba
What the Qwen API costs, model by model
Short answer. Alibaba lists 7 models with a per-token price on its API. Input: from $0.03 (Qwen3.7-Flash) to $2.50 (Qwen3.7-Max) per 1M tokens. Output: from $0.13 (Qwen3.7-Flash) to $7.50 (Qwen3.7-Max). Read on Alibaba's own pages between September 28 and November 16, 2026.
In plain words. You pay for the tokens you send (input) and the tokens the model writes (output). One request of 10,000 tokens in and 1,000 out costs $0.0004 on Qwen3.7-Flash and $0.0325 on Qwen3.7-Max. Output costs more than input on all 7 models.
Current models
7 models, cheapest input first, verified between September 28, 2026 and November 16, 2026. The last column is the cost of one request of 10,000 input and 1,000 output tokens at these prices, with no cache.
| Model | Input $/1M | Output $/1M | Cached input | Context | Example request | Verified |
|---|---|---|---|---|---|---|
| Qwen3.7-Flash | $0.03 | $0.13 | $0.006 | 1M | $0.0004 | 2026-10-28 |
| Qwen3.8-Flash | $0.15 | $0.47 | $0.016 | 1M | $0.0020 | 2026-10-28 |
| Qwen3.8-Omni-Flash | $0.15 | $0.47 | $0.016 | 1M | $0.0020 | 2026-09-28 |
| Qwen3.7-Plus | $0.40 | $1.60 | $0.08 | 1M | $0.0056 | 2026-10-30 |
| Qwen4-Max | $1.50 | $4.50 | $0.075 | 1M | $0.0195 | 2026-11-16 |
| Qwen3.8-Max | $2.00 | $6.00 | $0.25 | 1M | $0.026 | 2026-11-16 |
| Qwen3.7-Max | $2.50 | $7.50 | $0.50 | 1M | $0.0325 | 2026-10-11 |
How Alibaba describes its models
- Qwen3.8-Flash:
Qwen3.8-Flash is the multimodal model from the Qwen family, combining powerful reasoning and generation with remarkable speed.
(alibabacloud.com, read November 3, 2026) - Qwen3.8-Omni-Flash:
A model for audio and video understanding and content analysis, with text, image, audio, and video input and text output.
(alibabacloud.com, read October 1, 2026) - Qwen4-Max:
Qwen4-Max is the next-generation flagship multimodal model in the Qwen family, combining text generation, image and video understanding, and deep reasoning.
(alibabacloud.com, read November 16, 2026)
How Alibaba prices its models
- 7 of the 7 models publish a cached-input price, at 5%, 10.7%, 12.5% or 20% of the input price.
- 2 models charge more for long prompts : Qwen3.7-Flash $0.10 / $0.40 for prompts over 32,000 tokens, Qwen3.7-Plus $1.20 / $4.80 for prompts over 256,000 tokens.
- Prices shown here are Alibaba's Singapore (International) prices; in Beijing and Alibaba's Global regions (Frankfurt, Virginia, Tokyo, Hong Kong) the same models cost: Qwen3.7-Flash $0.028 / $0.11, Qwen3.8-Flash $0.113 / $0.382, Qwen3.8-Omni-Flash $0.113 / $0.382, Qwen3.7-Plus $0.276 / $1.101, Qwen4-Max $1.413 / $4.24, Qwen3.8-Max $1.65 / $4.951, Qwen3.7-Max $1.65 / $4.951.
Recorded price changes
- September 17, 2026 · added to this dataset, our omission. Omission on our side, not a price change by Alibaba. Qwen3.7 Flash belongs to the same generation as the Qwen3.7 Plus and Max rows and had no row in this dataset. It turned up on September 17, 2026 through the OpenRouter signal of AI Signal's read-only scan for missing models, and its prices were read that day on Alibaba's own Model Studio page: $0.03 input, $0.13 output and $0.006 cached per million tokens in the Singapore (International) region for prompts up to 32K tokens, $0.10 / $0.40 above 32K and $0.20 / $0.80 above 256K. For Alibaba the table covers the current Qwen generations, 3.7 and 3.8; the older generations Alibaba still sells, found the same day, were left out on purpose. source
- September 11, 2026 · change of our criterion. Change of criterion on our side, not a price change by Alibaba: no Alibaba price rose. Alibaba Cloud's Model Studio pages list the same model at two price levels: the China (Beijing) region and the Global regions (Frankfurt, Virginia, Tokyo, Hong Kong) share one level, and the Singapore region, labelled by Alibaba "Scope: International", is higher. Until September 11, 2026 the first-party table quoted the Beijing/Global level while the provider comparison quoted Alibaba's OpenRouter endpoint at the Singapore level, so the site showed two different "Alibaba" prices for the same model. From September 11 the reference region is Singapore (International), the one Alibaba labels international, the one every reseller we track (OpenRouter, Together, Fireworks) mirrors, and the higher of the two; the Beijing/Global figures stay published next to it. Published figures move from 1.65 / 4.951 to 2.00 / 6.00 (Qwen3.8 Max), 0.113 / 0.382 to 0.15 / 0.47 (Qwen3.8 Flash), 0.276 / 1.101 to 0.40 / 1.60 (Qwen3.7 Plus, prompts up to 256K; the tier above 256K, 1.20 / 4.80, is now recorded too) and 1.65 / 4.951 to 2.50 / 7.50 (Qwen3.7 Max). Alibaba's pages state that they show the original pricing excluding any limited-time promotions. source
Analysis that cites Alibaba models
- Prompt Caching Is Not 10% Everywhere: What a Cache Hit Costs Across Eight Providers · names 5 Alibaba models
- OpenAI Shuts Down GPT-4, GPT-4 Turbo, GPT-3.5 Turbo, o1, o1-pro and o3-mini on October 23, 2026: the Replacements and What They Cost · names 3 Alibaba models
- Qwen4-Max: $1.50 / $4.50 per Million Tokens, Under Qwen3.8-Max, With Cached Input 3.3× Cheaper · names 2 Alibaba models
- Five Things Decide What You Pay for a Model, and Only One of Them Is the Model · names 2 Alibaba models
- Mistral Large 4 Price Cut: $0.70 In, $2.10 Out, Still in Public Preview · names 1 Alibaba model
- Claude Haiku 5.5 Charges Five Times More Above 100,000 Tokens: What One Request Costs on Each Side of the Line · names 1 Alibaba model
- Claude Sonnet 4.5 Retires on November 30, 2026: What Moving to Claude Sonnet 5.5 Does to the Bill · names 1 Alibaba model
- Same List Price, Different Bill: Five Pairs of LLM APIs Where the Workload Decides What You Pay · names 1 Alibaba model
FAQ
What is the cheapest Qwen model on the API?
Qwen3.7-Flash, at $0.03 per 1M input tokens and $0.13 per 1M output tokens, as Alibaba's pricing page printed it on October 28, 2026.
How we checked
- Each price was read on Alibaba's pricing page (alibabacloud.com) or on a model page it links. The Verified column links to that page and gives the day.
- Quotes are copied as Alibaba's pages print them, with the day they were read. A script checks every week that the model descriptions are still on their pages.
- The example request is worked out from the per-1M prices: 10,000 input tokens and 1,000 output tokens, no cache, no batch discount, no long-prompt tier.
Full method: methodology. All providers: the full pricing table.