Qwen: Qwen3.5 397B A17B

Image input Text input Video input Text output
Author's Description

The Qwen3.5 series 397B-A17B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. It delivers state-of-the-art performance comparable to leading-edge models across a wide range of tasks, including language understanding, logical reasoning, code generation, agent-based tasks, image understanding, video understanding, and graphical user interface (GUI) interactions. With its robust code-generation and agent capabilities, the model exhibits strong generalization across diverse agent.

Key Specifications
Cost
$$$$$
Context
262K
Parameters
397B
Released
Feb 15, 2026
Speed
Ability
Reliability
Supported Parameters

This model supports the following parameters:

Presence Penalty Tools Response Format Temperature Top P Tool Choice Max Tokens Seed Reasoning Include Reasoning
Features

This model supports the following features:

Reasoning Tools Response Format
Performance Summary

The Qwen3.5 397B-A17B model, released by qwen on February 15, 2026, demonstrates exceptional reliability with a 100% success rate across all benchmarks, indicating consistent and dependable performance. However, it tends to have longer response times, ranking in the 6th percentile for speed, and is positioned at premium pricing levels, ranking in the 4th percentile for cost. Despite its speed and cost profile, the model exhibits strong performance across various tasks. It achieves near-perfect accuracy in Hallucinations (98.0%), showcasing its ability to appropriately acknowledge uncertainty. In Coding, it scores 92.9% accuracy, placing it in the 79th percentile. The model achieves perfect 100% accuracy in both Email Classification and Ethics, with Email Classification also being noted as the most accurate model at its price point and among models of its speed. Its Reasoning capabilities are also robust, with 96.0% accuracy, placing it in the 85th percentile. Key strengths include its outstanding reliability, perfect accuracy in critical classification and ethical reasoning tasks, and strong performance in coding and logical reasoning. Its primary weakness lies in its slower response times and premium pricing.

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $0.6
Completion $3.6

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
Alibaba
Alibaba | qwen/qwen3.5-397b-a17b-20260216 262K $0.6 / 1M tokens $3.6 / 1M tokens
AtlasCloud
AtlasCloud | qwen/qwen3.5-397b-a17b-20260216 262K $0.15 / 1M tokens $1 / 1M tokens
Novita
Novita | qwen/qwen3.5-397b-a17b-20260216 262K $0.6 / 1M tokens $3.6 / 1M tokens
Parasail
Parasail | qwen/qwen3.5-397b-a17b-20260216 262K $0.6 / 1M tokens $3.6 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Strategy Free Executions Accuracy Cost Duration
Other Models by qwen