Qwen 3.5 9B
qwen-3.5-9bqwen3-5-9bqwen3.5-9bqwen35-9b- 🆕 9B dense model in Alibaba's Qwen 3.5 series
- 🔧 Hybrid Gated DeltaNet attention architecture for efficient long context
- 📏 262K native context per the model card, extendable toward 1M
- 🌐 Supports 201 languages
- 🧠 Thinking/reasoning mode for step-by-step problem solving
- 👁️ Accepts text and image inputs
- 🔧 Native function calling for agentic workflows
- 🔒 Apache 2.0 licensed, open weights
Alibaba Group is a Chinese multinational technology company founded in 1999 and headquartered in Hangzhou, Zhejiang. Originally built around e-commerce and cloud computing, Alibaba has become one of the most prolific contributors to open-weight AI research, developing the Qwen…
Explore 39 more models by Alibaba Group →Qwen 3.5 9B is a compact model from Alibaba's Qwen team, part of the Qwen 3.5 series and published under the Apache 2.0 license in early 2026. It is built around a hybrid architecture that pairs Gated DeltaNet linear attention with full-attention layers, a design aimed at efficient processing across very long inputs. The model card lists a 262K native context window, extendable toward roughly 1M tokens, alongside support for 201 languages.
Within the same family, Qwen 3.5 9B sits below the mixture-of-experts releases Qwen 3.5 35B A3B and Qwen 3.5 397B, offering a small dense alternative for users who prefer a single-path model. Compared with the earlier dense Qwen 3.6 27B and prior Qwen generations, the 9B variant trades raw scale for a lighter footprint while retaining the series' core features.
Functionally, the model includes a thinking mode for explicit reasoning traces, native function calling for tool use and agentic pipelines, and image input support. It is distributed in FP8 in this catalog, and community quantizations such as 4-bit GGUF builds make local deployment more accessible. As with other open-weight Qwen releases, the weights and model card are hosted directly by the Qwen team on Hugging Face.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
| Seller | Reputation↓ | Routing | Input $/M | Cached $/M | Output $/M | Categories | API |
|---|---|---|---|---|---|---|---|
| Venice.ai Proxy 0x1f22…18c9 | 100.00 | gated | $0.05 | $0.05 | $0.075 | chat,reasoning,vision,multimodal,web-search | openai-chat-completions |
| Open Ant 0xe4f6…5bc4 | 74.05 | #1 | $0.10 | $0.0325 | $0.15 | chat,reasoning,vision,multimodal,tools,cheap | openai-chat-completions |
| Fire Ant 🔥🐜 0xbe05…bc5d | 41.29 | gated | $0.0791 | $0.0791 | $0.1187 | — | — |
| Open Bird 0xc0f1…8183 | 18.32 | gated | $0.05 | $0.05 | $0.075 | chat,open-source | openai-chat-completions |
| Meridian AI 0x8c8c…06f5 | 5.64 | gated | $0.0107 | $0.0107 | $0.0322 | chat,coding | openai-chat-completions |
| Skeffo Inference 0x1af8…e2b5 | 0.02 | gated | $0.05 | $0.05 | $0.15 | chat,vision,function-calling | openai-chat-completions |
| antseed-neon-puma-944e 0x6650…944e | 0.00 | gated | $0.0341 | $0.0341 | $0.0512 | chat,math | openai-chat-completions |
| Leftermute 0x388b…5389 | 0.00 | gated | $0.0045 | $0.0045 | $0.0136 | chat,coding,json | openai-chat-completions |
"Best price" and the seller table are live AntSeed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = on-chain trust (0-100). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on AntSeed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.