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 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.
Model card on Hugging Face- Input / output price
- $0.172 / $1.032 per 1M
- Context
- 262K tokens
- Released
- Feb 16, 2026
Paste into Claude Code, Cursor, or Codex
Help me set up the Respan gateway: fetch and follow https:/
Providers
9 providers, cheapest first, retired then OpenRouter last, priced per 1M tokens after gateway discounts.
| Provider | Input | Output |
|---|---|---|
Concentrate concentrate/qwen3.5-397b-a17b | $0.172/M | $1.032/M |
EmpirioLabs empiriolabs/qwen3.5-397b-a17b | $0.172/M | $1.032/M |
Chutes chutes/qwen3.5-397b-a17b | $0.45/M | $3.00/M |
DeepInfra deepinfra/qwen3.5-397b-a17b | $0.45/M | $3.00/M |
Parasail parasail/qwen3.5-397b-a17b | $0.50/M | $3.60/M |
Novita novita/qwen3.5-397b-a17b | $0.60/M | $3.60/M |
Qwen qwen/qwen3.5-397b-a17b | $0.60/M | $3.60/M |
NebiusRetired nebius/qwen3.5-397b-a17b | $0.60/M | $3.60/M |
OpenRouter openrouter/qwen3.5-397b-a17b | $0.45/M | $3.00/M |
Is a provider down? OpenRouter status
Pricing
Averaged across 9 providers, in USD per 1M tokens after gateway discounts.
- Average input
- $0.44 per 1M
- Average output
- $2.83 per 1M
- Average cache read
- $0.18 per 1M
Cost by volume
Total cost from 0 to 1T tokens on the cheapest and most expensive provider.
Throughput and latency
30-day averages from requests through the Respan gateway. Under each figure, every model on the gateway by its average, this one in blue.
Output speed
Higher is better24tokens/sacross providers
Response latency
Lower is better3.7sacross providers
Benchmarks
The Respan Index plus 19 published scores from 8 sources, each linked to where it was published.
29.5/ 100
Respan Index#93 of 175
- 93Qwen3.5 397B-A17B29.5
- 92Gemini 3.5 Flash-Lite29.9
- 94Inkling-Small29.4
Published scores
| Benchmark | Score | Source | Published |
|---|---|---|---|
| SWE-bench VerifiedQwen | 76.4% | Qwen | Feb 15, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| tau2-bench TelecomSierra | 97.8% | Sierra | Mar 2, 2026 |
| tau2-bench RetailSierra | 84.4% | Sierra | Mar 2, 2026 |
| tau2-bench AirlineSierra | 81.5% | Sierra | Mar 2, 2026 |
| Toolathlon-VerifiedHKUST | 40.7% | HKUST | Jul 30, 2026 |
| APEX-AgentsMercor | 24.9% | Mercor | Oct 5, 2026 |
| tau2-bench Banking KnowledgeSierra | 9.8% | Sierra | Mar 2, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| GPQA DiamondQwen | 88.4% | Qwen | Apr 21, 2026 |
| GPQA DiamondEpoch AI | 86.4% | Epoch AI | Oct 5, 2026 |
| Mystery Game PuzzlesEpoch AI | 18% | Epoch AI | Oct 5, 2026 |
| Chess PuzzlesEpoch AI | 13% | Epoch AI | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| AIME 2026MathArena | 94.2% | MathArena | Sep 17, 2026 |
| OTIS Mock AIME 2024-2025Epoch AI | 88.9% | Epoch AI | Oct 5, 2026 |
| FrontierMath Tiers 1-3 v2Epoch AI | 31.2% | Epoch AI | Oct 5, 2026 |
| Benchmark | Score | Source | Published |
|---|---|---|---|
| LMArena EloLMArena | 1441 Elo · #72 of 187 | LMArena | Nov 18, 2026 |
| LMArena WebDevLMArena | 1400 Elo · #87 of 123 | LMArena | Nov 18, 2026 |
| LMArena VisionLMArena | 1246 Elo · #47 of 89 | LMArena | Nov 9, 2026 |
How Qwen compares: DeepSeek vs Qwen, Claude vs Qwen, Gemini vs Qwen
Compare Qwen3.5 397B A17B: Qwen3.5 397B-A17B vs Qwen3.8-Max, Qwen3.5 397B-A17B vs Qwen3.8-Flash-Next, Qwen3.5 397B-A17B vs Qwen3.7-Max, Qwen3.5 397B-A17B vs Qwen3.8 27B, Qwen3.5 397B-A17B vs Qwen3.6-Max-Preview, Qwen3.5 397B-A17B vs Qwen3.7-Plus
Apps
Use Qwen3.5 397B A17B in the tools you already run. Each guide points the app at the Respan gateway.
Setup guides for more apps are in the integration docs.
Qwen3.5 397B A17B FAQ
Short answers, built from the data on this page.
Qwen3.5 397B A17B is a model from Qwen, released on Feb 16, 2026. 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.
On the Respan gateway, Qwen3.5 397B A17B costs from $0.172 per 1M input tokens and $1.032 per 1M output tokens on Concentrate and EmpirioLabs, after gateway discounts. Averaged across its 9 providers, it costs $0.44 per 1M input tokens and $2.83 per 1M output tokens.
Qwen3.5 397B A17B is available through 9 providers: Concentrate, EmpirioLabs, Chutes, DeepInfra, Parasail, Novita, and 3 more.
Qwen3.5 397B A17B accepts up to 262K tokens of context on its largest provider route.
Point any OpenAI-compatible SDK at https://gateway.respan.ai/v1 with your Respan API key and set the model to "qwen/qwen3.5-397b-a17b". To set it up with a coding agent, copy the agent setup prompt at the top of this page.
Qwen3.5 397B A17B scores 29.5 out of 100 on the Respan Index v1.1, ranking #93 of 175 models. By category: Coding 11.2, Agents 34.3, Reasoning 30.8, Knowledge & language 27.6, and Human preference 51.1. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by Respan.
Qwen3.5 397B A17B has 19 published benchmark scores, including SWE-bench Verified 76.4%, tau2-bench Telecom 97.8%, and GPQA Diamond 88.4%. Each score links to where it was published; the data is collected by Respan.
Certified where it counts.
Built to meet the security and privacy standards that enterprise and healthcare teams require.
ISO 27001
The internationally recognized standard for information security management.
SOC 2
Secure, compliant management of your data across all of our systems.
GDPR
Operated under GDPR, the world's strictest standard for data privacy.
HIPAA
HIPAA compliant, with a BAA available for healthcare teams.