Qwen3.5 397B is the open-weight Qwen3.5-397B-A17B model from Qwen, with 397B total and 17B activated parameters, released in February 2026. the weights accept images and video, but this face is served text-only.
pick it if
want Apache 2.0 open weights you could later self-host
want a thinking-by-default model with tool calling
reuse long prompts: cache hits bill at $0.35 versus $0.60 input
skip it if
you need image input: Qwen3.7 Plus takes images and costs less
your workload is output-heavy and a cheaper sibling covers it
Qwen3.7 Plus is Alibaba's multimodal agent model from the Qwen3.7 generation, served through Model Studio with text, image and video input. in this catalog it sits between Qwen3.8 Omni Flash and Qwen3.8 Max on price.
pick it if
you want a Qwen model with image and video input below Qwen3.8 Max pricing
you reuse long prompts: cache hits bill at a fifth of input
you run screen, GUI or tool-using agents, the work Qwen pitches it for
skip it if
you need audio input; Qwen3.8 Omni Flash takes it for less
you need more than the 256K context window this catalog serves
faq
which is cheaper, Qwen3.5 397B or Qwen3.7 Plus?
Qwen3.7 Plus lists lower input and output prices. on a coding-agent session (2M tokens in, 60,000 out) Qwen3.5 397B costs $1.42 and Qwen3.7 Plus $0.90.
which one accepts images?
Qwen3.7 Plus supports vision; Qwen3.5 397B does not.
can i switch between Qwen3.5 397B and Qwen3.7 Plus mid-task?
yes. in an octomind session, /model octohub:qwen3.5-397b and /model octohub:qwen3.7-plus switch models without losing the conversation, so you can draft on one and review on the other.
try either model
switch between models inside the same octomind session.