ZMIME
Alibaba (Qwen) · Released Feb 15, 2026

Qwen3.5 397B-A17B

Large open Qwen multimodal MoE for visual agents and long technical tasks.

  • Open weights
  • Reasoning
  • Tool calling
  • Vision
  • Audio input
  • Video input
Capability
146.7ECI · #67 of 148
Input
$0.60per 1M tokens
Output
$3.60per 1M tokens
Context
262K66K max output
Pricing

What it costs

Input
$0.60per million tokens
Output
$3.60per million tokens
Blended (3:1)
$1.35pricier than 62% of priced models

Typical monthly bills

Monthly usageEstimated cost
Side project2M input + 0.5M output tokens$3.00
Team assistant25M input + 5M output tokens$33.00
Production app250M input + 50M output tokens$330.00

Official Alibaba API price, as listed on models.dev.

Capability

Independent benchmarks

146.7Epoch Capabilities Index
#67 of 148 scored models
88Median 146167

The shaded band is Epoch AI’s confidence range (144.8–148.2); the tick marks the median scored model.

  • GPQA DiamondGraduate-level science questions86.4%
  • FrontierMath Tiers 1–3Research-level mathematics31.2%
  • OTIS Mock AIME 2024–2025Competition mathematics88.9%

Scores from Epoch AI, run independently of Alibaba (Qwen).

Specifications

The details

Lab
Alibaba (Qwen)
Released
Feb 15, 2026
Context window
262,144 tokens
Max output
65,536 tokens
Inputs
Text, Images, Audio, Video
Output
Text
Reasoning
Thinking budget
Tool calling
Yes
Structured output
Yes
API model ID
qwen3.5-397b-a17bon Alibaba
Availability
23 API providerslisted on models.dev

Alibaba model documentation

Lab-reported

What Alibaba (Qwen) claims

Published by the lab at launch. Settings vary, so compare these only with care.

BenchmarkScoreSettingSource
SWE-Bench Verified76.4 resolved—Source
Same lab

More from Alibaba (Qwen)

Questions

About Qwen3.5 397B-A17B

How much does Qwen3.5 397B-A17B cost?

Qwen3.5 397B-A17B costs $0.60 input / $3.60 output per million tokens (official Alibaba API price). At a 3:1 input-to-output mix that is $1.35 per million tokens, more expensive than 62% of the 360 priced models we track.

What is the context window of Qwen3.5 397B-A17B?

Qwen3.5 397B-A17B accepts up to 262,144 tokens per request and can write up to 65,536 tokens in one response.

How good is Qwen3.5 397B-A17B?

Epoch AI gives Qwen3.5 397B-A17B a Capabilities Index score of 146.7 (likely range 144.8–148.2), ranking it #67 of 148 models Epoch has scored. Epoch AI benchmark results: GPQA Diamond 86.4%, FrontierMath Tiers 1–3 31.2%, OTIS Mock AIME 2024–2025 88.9%.

Is Qwen3.5 397B-A17B open source?

Yes. Alibaba (Qwen) publishes the weights, so it can be downloaded and self-hosted.

What inputs does Qwen3.5 397B-A17B support?

Qwen3.5 397B-A17B accepts text, images, audio and video and replies in text. It is a reasoning model, supports tool calling and can return structured JSON output.

When was Qwen3.5 397B-A17B released?

Alibaba (Qwen) released Qwen3.5 397B-A17B on Feb 15, 2026.

What are the best alternatives to Qwen3.5 397B-A17B?

The closest current models from other labs on capability, price and release date are GLM-5 (Z.ai (Zhipu), ECI 145.8, $1.00 / $3.20), Kimi K2.5 (Moonshot AI, ECI 148.0, $0.60 / $3.00), GPT-5.4 mini (OpenAI, ECI 148.8, $0.75 / $4.50) and Nemotron 3 Ultra 550B A55B (NVIDIA, ECI 146.2, $0.50 / $2.50).