QwenQwen 3.5 9B
Qwen makes Qwen 3.5 9B, an open-source chat model that accepts text and image inputs and returns text outputs. It is designed for conversational tasks and can call tools for web search, code execution, and function calling through its API, and supports structured outputs and extended reasoning.
Input
Output
Context
262K
Max Output
33K
Parameters
9.7B
Technical Specifications
Input Modalities
Output Modalities
Some values come from web research and may be incorrect. Report incorrect data
Capabilities
Strengths
- Open weights under a permissive license (apache-2.0)
- 262K token context window
- Runs on one 16 GB GPU at INT8 (NVIDIA T4)
Limitations
- Below average on long-context reasoning among measured models
Benchmarks
Measured speed: 80 tok/s output speed · 0.48 s to first answer token · 0.48 s to first token
Bands compare this model with every model the same suite measured in our catalog.
Popular comparisons
Compare with any model →- Qwen 3.5 9B vs Qwen 3.5 4BSame family · 4.7B parameters
- Qwen 3.5 9B vs Qwen 3.5 27BSame family · 28B parameters
- Qwen 3.5 9B vs Step 3 VL 10BSimilar size · 10B parameters
- Qwen 3.5 9B vs Qwen 3 VL 8B InstructSimilar size · 8.8B parameters
- Qwen 3.5 9B vs Llama 3.2 11B Vision InstructSimilar size · 11B parameters
- Qwen 3.5 9B vs Llama Guard 4 12BSimilar size · 12B parameters
Estimated hardware requirements
Size it in the GPU planner →Download 19.3 GB · Runs on vLLM v0.32.0 · 354 GGUF builds
| Precision | Weights | One request at 33K tokens | Smallest GPU that fits |
|---|---|---|---|
| FP16 / BF16 | 18.0 GB | 22.7 GB | 1x Tenstorrent Blackhole p150 (32 GB) |
| INT8 | 8.99 GB | 13.2 GB | 1x NVIDIA T4 (16 GB) |
| INT4 | 4.50 GB | 8.52 GB | 1x NVIDIA T4 (16 GB) |
- Quantized builds
- GGUF 354 · MLX 114 · AWQ 11 · GPTQ 22
- Runtimes
- vLLM v0.32.0 · llama.cpp, Ollama, LM Studio (GGUF) · MLX on Apple silicon
Computed by Inferbase from the 9,653,104,368 parameters in the creator's Hugging Face repository; one request adds its KV cache, activations and framework overhead, with the layer layout estimated from the parameter count.
Resources & Links
Public resources
Specification sources
- Technical specifications
- Hugging Face
- Max output, training cutoff
- Web research
- Context window
- Creator or provider API
As of Nov 14, 2026.
Frequently asked questions about Qwen 3.5 9B
Answered from the facts on this page, with their sources.
Qwen 3.5 9B has a context window of 262,144 tokens and a maximum output of 32,768 tokens (source: Creator or provider API). A provider may serve less; Inferbase routes on what each provider serves.
Yes. Qwen 3.5 9B is an open-weight model under the Apache 2.0 license. You can download and self-host it within the license terms.
Qwen 3.5 9B has 9,653,104,368 parameters (9.7B), as stated in the creator's Hugging Face repository.
Strengths: Open weights under a permissive license (apache-2.0); 262K token context window. Limitations: Below average on long-context reasoning among measured models. Each is computed from the model's stated facts and benchmark results.
Qwen 3.5 9B accepts Text and Image and produces Text. The Capabilities card lists the API features it supports, such as function calling and structured output.
Qwen 3.5 9B has published results from Artificial Analysis and Epoch AI. The Benchmarks card shows each score as reported, with its band among the models the same source measured. Scores are never combined across sources.
Yes. Qwen 3.5 9B is available on Inferbase Managed through the OpenAI-compatible API as qwen3-5-9b. Request it by name, or request auto and routing picks it when it fits the request best.
The weights take 18.0 GB at FP16 / BF16, 9.0 GB at INT8, 4.5 GB at INT4, computed from the 9,653,104,368 parameters its creator publishes. Each request adds its KV cache and the framework's overhead. The hardware section shows one request and the smallest GPU that fits; the GPU planner sizes it for your context length and traffic.
Put intelligence in the middle.
Start on the free plan with 5,000 requests a month, no card required.