Skip to content
RankShark

Qwen3.5 397B A17B

The Qwen3.5 series 397B-A17B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism wit

ModelOpen sourceby Qwensince 2026
38.0

SharkScore

#150 in Models

Why it scores 38.0

Qwen3.5 397B A17B has a SharkScore of 38.0/100, ranked #150 of 608 in Models (updated 2026-11-19).

Each part compares Qwen3.5 397B A17B with other listings in Models. Evidence covers 87% of the score's weight.

Adoption
31 · 30%
User reviews
no data · 25%
Maintenance
11 · 20%
Trust & transparency
80 · 15%
Momentum
no data · 10%

Paying never changes this score. How ranking works · method v2026-11-18.2

RankShark review of Qwen3.5 397B A17B

Automated evaluation by RankShark · 18 Nov 2026 · method v1.0 · How we review

We were not paid for this review.

As of 2026-11-18, Qwen3.5 397B A17B is an open-weight vision-language model; OpenRouter lists usage at $0.45 per million input tokens and $3 per million output tokens. [E1][E2]

Qwen3.5 397B A17B is a vision-language model with 397 billion total parameters and 17 billion activated parameters. Its weights are available in Hugging Face Transformers format, and OpenRouter lists text, image, and video inputs with text output. [E1][E2]

Best for

  • Developers looking to run model weights with supported inference frameworks. · model card
  • Applications that need text, image, or video input and text output. · OpenRouter
  • Long-context workloads, with a listed context length of 262,144 tokens. · model card, OpenRouter

Not for

  • Applications that require the model to generate images or video; OpenRouter lists text as the output modality. · OpenRouter

Key features

  • Vision-language inputs. The vendor describes the model as a causal language model with a vision encoder; OpenRouter lists text, image, and video inputs and text output. · model card, OpenRouter
  • Hybrid architecture. The vendor says the model combines Gated Delta Networks with sparse Mixture-of-Experts. · model card
  • Model size. The vendor lists 397B total parameters and 17B activated parameters. · model card
  • Long context. The vendor lists 262,144 tokens natively and says context can be extended up to 1,010,000 tokens; OpenRouter lists 262,144 tokens. · model card, OpenRouter
  • Model weights and framework compatibility. The vendor says the repository contains model weights and configuration files in Hugging Face Transformers format, compatible with Transformers, vLLM, SGLang, and KTransformers. · model card
  • OpenRouter API endpoints. OpenRouter lists chat completions, completions, messages, and responses endpoints. · OpenRouter

Pros

  • Weights are available for use with several named inference frameworks. · model card
  • The listed inputs include text, images, and video. · OpenRouter
  • The vendor lists a native context length of 262,144 tokens. · model card

Cons

  • OpenRouter lists text as the output modality, not image or video generation. · OpenRouter
  • OpenRouter lists a maximum completion of 81,920 tokens, below its 262,144-token context length. · OpenRouter
  • No free plan or free trial stated on the pages we checked.
  • Whether customer data is used for model training is not stated on the pages we checked.

Pricing

PlanPriceLimits
OpenRouter model listing$0.45 per million input tokens; $3 per million output tokensOpenRouter lists a 262,144-token context length and an 81,920-token maximum completion.

Checked 18 Nov 2026

Free plan
Not stated (checked 18 Nov 2026)
Free trial
Not stated (checked 18 Nov 2026)
Trains on customer data
Not stated (checked 18 Nov 2026)
Data retention
Not stated (checked 18 Nov 2026)
Data regions
Not stated (checked 18 Nov 2026)
Data processing agreement
Not stated (checked 18 Nov 2026)
AI models used
Qwen3.5-397B-A17B; OpenRouter describes it as a native vision-language model. · model card, OpenRouter
API
OpenRouter lists chat completions, completions, messages, and responses endpoints. · OpenRouter
SDKs
Not stated (checked 18 Nov 2026)
Webhooks
Not stated (checked 18 Nov 2026)
Platforms
The model repository is in Hugging Face Transformers format and is listed as compatible with Transformers, vLLM, SGLang, and KTransformers. The vendor also points to Alibaba Cloud Model Studio for hosted inference. · model card
Security claims
No security certification stated (checked 18 Nov 2026)

What we measured

Context window (OpenRouter listing) · now
262,144 tokens
Price per million tokens (OpenRouter listing) · now
$0.45 in, $3 out
Package downloads · 30 days
711,939
How we tested · sources
  • Hourly uptime and response-time checks of the product's URL.
  • Repository, release and download data from GitHub, npm, PyPI and Hugging Face where they exist.
  • We did not sign up, pay, or use the product as a customer.
  • 18 Nov 2026: Plans or prices changed on the pricing page.
  • 18 Nov 2026: Plans or prices changed on the pricing page.

Something wrong? Report an error with a public link that shows it; corrections re-run the review.

Top alternatives to Qwen3.5 397B A17B

See all
  • 67.1

    Qwen3.8 27B is an open-weight dense vision-language model from Qwen.

    ModelOpen source
  • 65.5

    Qwen3.6-35B-A3B is an open-weight multimodal model from Alibaba Cloud with 35 billion total parameters and 3 billion active parameters per t

    ModelOpen source
  • 64.7

    Qwen3.8 Flash is a multimodal reasoning model from Alibaba.

    ModelOpen source
  • 64.3

    Qwen3.6 27B is a dense 27-billion-parameter language model from the Qwen Team at Alibaba, released in April 2026.

    ModelOpen source