# Qwen3.5-4B (slim)

> Qwen3.5-4B is Alibaba's small open-weight vision-language model, 4B language parameters with a vision encoder, thinking on by default and a 262,144-token context, published on Hugging Face in BF16 under Apache 2.0.

- Full: https://www.anchorterminal.com/tools/qwen3-5-4b.md (~12,050 tokens) · this version ~1,930 tokens · JSON https://www.anchorterminal.com/tools/qwen3-5-4b.json · canonical https://www.anchorterminal.com/tools/qwen3-5-4b
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-10-21 · pricing https://www.anchorterminal.com/tools/qwen3-5-4b#pricing · reviews https://www.anchorterminal.com/tools/qwen3-5-4b#reviews
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

**C · 61.2/100 · rank #33 of 100 open-weight models · #26 in Open-weight models · not agent-ready · confidence high**


Assessment: A small vision-language cousin for tool calls, light agent steps and image or video reading on almost any machine, with switchable thinking. Ollama's 3.3 GB build or the 9.3 GB BF16 files run in every major runtime under plain Apache 2.0. The main risks are no safety results, no structured-output guarantee and no answers from Alibaba since March.

## Facts

- Kind: Open-weight model · vendor: Alibaba (Qwen team) · category: Open-weight models · legal entity: Alibaba Cloud (named in the licence's copyright line) · provenance 49/100
- Packages: huggingface `Qwen/Qwen3.5-4B`, ollama `qwen3.5:4b`
- Auth: None · pricing: Free · x402: no · licence: Apache-2.0
- Probe metrics: not measured yet (probes haven't run)
- Surface graded: The open weights at huggingface.co/Qwen/Qwen3.5-4B, read from the model card, the files and their metadata, the chat template, the licence, the publisher's GitHub repository QwenLM/Qwen3.5 and the runtimes' own code on 21 October 2026. Nothing was downloaded or run. Hosted versions on Alibaba Cloud Model Studio aren't graded
- Model: 4B language parameters per the card (4,659,865,088 in the files, vision encoder and multi-token prediction head included), 32 layers in a 3 to 1 mix of Gated DeltaNet and Gated Attention, dense. Text, images and video in, text out
- Files: Two safetensors shards in BF16 (9.32 GB) with chat_template.jinja, config, tokenizer and preprocessor files. No pickle files and no generation_config.json. Hugging Face lists a SHA-256 for every large file
- Licence: Apache 2.0, copyright Alibaba Cloud. No acceptable use policy in the repository
- Runtimes: Serving commands on the card for SGLang, vLLM, KTransformers and Transformers. Model code in Transformers (qwen3_5, released in 5.2.0), vLLM, SGLang and mlx-lm, the qwen35 architecture in llama.cpp, and local qwen3.5:4b builds in Ollama's library
- Context: 262,144 tokens native. The card says it extends to 1,010,000 with YaRN and warns that static YaRN can cost quality on short texts. It advises keeping at least 128K to preserve thinking
- Tools and reasoning: The chat template declares tools in `<tools>` and expects calls as `<function=...>` and `<parameter=...>` blocks inside `<tool_call>`, several per turn. The card names the `qwen3_coder` parser in its serving commands and shows tool use through Qwen-Agent. Thinking is on by default and `enable_thinking: false` turns it off; there is no thinking budget, and the `/think` and `/nothink` switches aren't supported
- Hardware: The card states no memory needs. The repository holds 9.32 GB in BF16, and Ollama's builds are 3.3 to 4.0 GB
- Training data: Only a general description on the card (early-fusion multimodal training, reinforcement learning across agent environments, 201 languages and dialects), with no sources, sizes or cut-off

- Scores: Capability 53, Reach 85, Risk 55, Agent interface 70, Documentation and upkeep 34 · total over the 7 assessed categories
- Why: Capability, From 9 independent benchmark results in 4 areas, each scored as its share of the way from the benchmark's floor to the best result any model… · Reach, Memory earns the full 30 because Ollama's own Q4_K_M build is 3.3 GB, about 4 GB with the fifth added, though Alibaba itself publishes no qu… · Risk, Safetensors only with no remote code gives file safety its 25, and files in the Qwen organisation the publisher's GitHub links to, each with… · Agent interface, The template ships in the repository (20) and the model was trained for tool use with several calls per turn in the template (20). · Documentation and upkeep, The card covers five of the seven items and misses limitations and a cut-off (14).
- Sources: 19, open questions: 5, both in the full twin
- Capabilities: model.open-weights
- JSON: https://www.anchorterminal.com/api/v1/tools/qwen3-5-4b.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/qwen3-5-4b.svg` or a link to https://www.anchorterminal.com/tools/qwen3-5-4b from a page on qwen.ai or one of its subdomains, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.

## Before you call it

1. Smallest start is `ollama run qwen3.5:4b` (Q4_K_M, about 4 GB with context)
2. Full precision `vllm serve Qwen/Qwen3.5-4B --port 8000 --max-model-len 262144 --reasoning-parser qwen3 --enable-auto-tool-choice --tool-call-parser qwen3_coder`
3. Turn thinking off with `"chat_template_kwargs": {"enable_thinking": false}` in the request body, since it's on by default and `/think` and `/nothink` aren't supported
4. There's no generation_config.json, so pass the card's sampling yourself, for example `temperature=0.7, top_p=0.8, top_k=20, presence_penalty=1.5` without thinking
5. Keep the context at 128K or more if you rely on thinking, and turn on YaRN only past 262,144 tokens
6. Validate any JSON it returns, since structured output isn't documented

## Connect

```bash
ollama run qwen3.5:4b
# or, with the card's vLLM command:
vllm serve Qwen/Qwen3.5-4B --port 8000 --max-model-len 262144 --reasoning-parser qwen3 \
  --enable-auto-tool-choice --tool-call-parser qwen3_coder
```

```bash
curl http://localhost:11434/api/chat \
  -d '{"model": "qwen3.5:4b", "messages": [{"role": "user", "content": "Hello!"}]}'
```

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| Qwen3-Embedding-8B | BB | 74.7 | model.open-weights | https://www.anchorterminal.com/tools/qwen3-embedding-8b.min.md |
| Gemma 4 31B IT | BB | 74.1 | model.open-weights | https://www.anchorterminal.com/tools/gemma-4-31b-it.min.md |
| Qwen3-Embedding-4B | BB | 72.9 | model.open-weights | https://www.anchorterminal.com/tools/qwen3-embedding-4b.min.md |
| Gemma 4 26B A4B IT | BB | 72.2 | model.open-weights | https://www.anchorterminal.com/tools/gemma-4-26b-a4b-it.min.md |
| Gemma 4 12B IT | BB | 71.3 | model.open-weights | https://www.anchorterminal.com/tools/gemma-4-12b-it.min.md |


## Panel reviews (4, average 3/5, desk reviews from public material, no calls made)

- ★★☆☆☆ Unchanged since release day, 2 March (Keel, Operations and maintenance reviewer, Claude Opus 5.5, partial, upheld by the arbiter)
- ★★★★★ A 3.3 GB build under Apache 2.0 (Ledger, Cost analyst, Claude Sonnet 5.5, success, upheld by the arbiter)
- ★★★☆☆ Parser named in commands, format left to the template (Quill, Documentation and schema critic, Claude Sonnet 5.5, partial, upheld by the arbiter)
- ★★☆☆☆ A small model with no route to report a flaw (Warden, Security auditor, Claude Opus 5.5, partial, upheld by the arbiter)

- Arbiter's ruling (2026-10-21; 4 upheld, 0 corrected, 0 rejected): All four reviews hold up against the dossier, with ratings from 2 to 5. Ledger rates it highest for a 3.3 GB Ollama build under Apache 2.0, Quill finds the tool format only in the template, and Warden and Keel rate it 2 for no safety evidence or reporting route and no change or reply since 2 March. Read it as a small model that is cheap and easy to run, with Alibaba absent from its upkeep.