---
id: "qwen/qwen3.5-397b-a17b"
also: ["qwen/qwen3.5-397b-a17b-20260216", "Qwen/Qwen3.5-397B-A17B", "hf/Qwen/Qwen3.5-397B-A17B", "requesty/alibaba/qwen3.5-397b-a17b"]
cheapest_list_usd_per_m: "0.39 in, 2.34 out, at alibaba on openrouter"
cheapest_effective_usd_per_m: "0.3211 in, 3.0 out, at deepinfra/fp8 on openrouter"
context: 262144
max_output: 64000
input: ["text", "image", "video"]
output: ["text"]
released: "2026-02-16"
api: ["/api/v1/chat/completions", "/api/v1/completions", "/api/v1/messages", "/api/v1/responses"]
params: ["frequency_penalty", "include_reasoning", "logit_bias", "logprobs", "max_tokens", "min_p", "presence_penalty", "reasoning", "repetition_penalty", "response_format", "seed", "stop", "structured_outputs", "temperature", "tool_choice", "tools", "top_k", "top_logprobs", "top_p"]
endpoints: 19
gateways: "openrouter (11), hf (5), requesty (3)"
no_training: 13
trains: 0
retains_prompts: 4
coding: 48.2000
agentic: 8.3000
---
# Qwen: Qwen3.5 397B A17B

Qwen3.5-397B-A17B is Alibaba's most capable Qwen3.5 model, a Mixture-of-Experts architecture with 397B total parameters and 17B activated per token. It features a 262K token context window (extensible to 1M with YaRN), thinking/reasoning mode, tool calling with MCP integration, and support for 201 languages. Sets state-of-the-art results on reasoning, coding, math, and multimodal benchmarks.