BeavBeav
2.1K
中文Sign in

Alibaba Cloud · Text

qwen3.7-flash

Qwen multimodal understanding and task execution

VisionCodingAgents

Pricing

Charged by actual usage

Unit: points / 1M tokens

Input per request≤ 32K

Input
65≈ ¥0.65Member price 39≈ ¥0.39
Cache hit
13≈ ¥0.13Member price 7.8≈ ¥0.078
Cache write
81≈ ¥0.81Member price 48.6≈ ¥0.486
Output
259≈ ¥2.59Member price 155≈ ¥1.55

Input per request32K–256K

Input
194≈ ¥1.94Member price 116≈ ¥1.16
Cache hit
39≈ ¥0.39Member price 23.4≈ ¥0.234
Cache write
242≈ ¥2.42Member price 145≈ ¥1.45
Output
775≈ ¥7.75Member price 465≈ ¥4.65

Input per request256K–1M

Input
388≈ ¥3.88Member price 233≈ ¥2.33
Cache hit
78≈ ¥0.78Member price 46.8≈ ¥0.468
Cache write
484≈ ¥4.84Member price 290≈ ¥2.9
Output
1,549≈ ¥15.49Member price 929≈ ¥9.29
  • Paid members are charged the member price, 40% below the list price.
  • Tiered pricing: the tier is picked by the input length of each request. Whichever tier a request’s input tokens fall into, its input, cache and output are all billed at that tier’s prices, not split across tiers.
  • Cache hit: a prefix repeated from earlier requests (such as a system prompt or a long document you keep referring to) is cached, and the cached part is billed at the cache price, about 20% of the input price.
  • Cache write: content written to the cache the first time is billed at the cache write price; later hits are billed at the cache hit price.
  • Input and output are billed separately by actual tokens; reasoning tokens count as output.

Lifetime and Team members get 40% off this model. See plans

About this model

Qwen3.7-Flash is a native vision-language model supporting text, image, and video inputs. It strengthens visual recognition, spatial understanding, multimodal coding, and agent task execution.

Use in Beav

  1. 01Download Beav and sign in
  2. 02Choose qwen3.7-flash in the model picker
  3. 03Use it for writing, editing, summarizing and chat

Similar models

View all