- Size
- 4.7B parameterscounted from the safetensors header on the Hub
- Licence
- custom, read it, Apache License 2.0Our reading, check the licence: A custom licence we have not read for this model: other makers use the same name for different texts. Read the model's own licence file before you use it.Licence text
- On the Hub since
- 27 Feb 2026the date the repo was created
- Context length
- 256k tokensfrom the model's config
- Fine-tune
- of Qwen/Qwen3.5-4B-BaseQwen3_5ForConditionalGeneration
- On Hugging Face
- Qwen/Qwen3.5-4B4.6M downloads in the last 30 days, 1.3k likes
Scores
- effort none55.8%±6.228 Aug 2026, log
Epoch's name: qwen3.5-4b. Best closed model on this set: claude-fable-5_high, 100.0% (Anthropic). All 318 results
- effort none0.0%via Epoch AI
Epoch's name: qwen3.5-4b. Best closed model on this set: gemini-4-carbon_high, 36.0% (Google DeepMind). All 271 results
Claimed by the maker Claimed
The model card has no claimed scores in a machine-readable form (the card's model-index), so none are shown. The maker may give scores in the card text: read the card.
Other views, which we link to and do not copy: look this model up on Artificial Analysis and the Arena leaderboard.
Independent scores from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Each row links its own source. What the badges mean.
Run it yourself
Memory needed = weights + context cache + 0.50 GB for the runtime. Context cache at 8k tokens: 0.25 GB, worked out from the model's config (Qwen/Qwen3.5-4B), at 16-bit precision. Choose a machine above to see a verdict for each format. Method
GGUF for llama.cpp, Ollama, LM Studio and others
unsloth/Qwen3.5-4B-GGUF (converted by unsloth), 1.7M downloads. Sizes are the real files on the Hub, all parts added up. For image input you also need the projector file mmproj-F16.gguf (0.63 GB), not counted below.
| Format | Size | Memory needed | Get it |
|---|---|---|---|
| UD-IQ3_XXS | 1.8 GBfile size | 2.6 GB1.8 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:UD-IQ3_XXSOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:UD-IQ3_XXSDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-UD-IQ3_XXS.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q3_K_S | 2.0 GBfile size | 2.7 GB2.0 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q3_K_SOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q3_K_SDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q3_K_S.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q3_K_M | 2.1 GBfile size | 2.9 GB2.1 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q3_K_MOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q3_K_MDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q3_K_M.gguf" --local-dir Qwen3.5-4B-GGUF |
| UD-Q3_K_XL | 2.3 GBfile size | 3.0 GB2.3 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:UD-Q3_K_XLOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:UD-Q3_K_XLDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-UD-Q3_K_XL.gguf" --local-dir Qwen3.5-4B-GGUF |
| IQ4_XS | 2.3 GBfile size | 3.1 GB2.3 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:IQ4_XSOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:IQ4_XSDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-IQ4_XS.gguf" --local-dir Qwen3.5-4B-GGUF |
| IQ4_NL | 2.4 GBfile size | 3.2 GB2.4 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:IQ4_NLOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:IQ4_NLDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-IQ4_NL.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q4_0 | 2.4 GBfile size | 3.2 GB2.4 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q4_0Ollama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_0Download hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q4_0.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q4_K_S | 2.4 GBfile size | 3.2 GB2.4 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q4_K_SOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_K_SDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q4_K_S.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q4_K_M | 2.6 GBfile size | 3.3 GB2.6 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q4_K_MOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_K_MDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q4_K_M.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q4_1 | 2.6 GBfile size | 3.3 GB2.6 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q4_1Ollama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_1Download hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q4_1.gguf" --local-dir Qwen3.5-4B-GGUF |
| UD-Q4_K_XL | 2.7 GBfile size | 3.5 GB2.7 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:UD-Q4_K_XLOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:UD-Q4_K_XLDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-UD-Q4_K_XL.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q5_K_S | 2.8 GBfile size | 3.6 GB2.8 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q5_K_SOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q5_K_SDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q5_K_S.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q5_K_M | 2.9 GBfile size | 3.7 GB2.9 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q5_K_MOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q5_K_MDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q5_K_M.gguf" --local-dir Qwen3.5-4B-GGUF |
| UD-Q5_K_XL | 3.0 GBfile size | 3.8 GB3.0 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:UD-Q5_K_XLOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:UD-Q5_K_XLDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-UD-Q5_K_XL.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q6_K | 3.3 GBfile size | 4.0 GB3.3 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q6_KOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q6_KDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q6_K.gguf" --local-dir Qwen3.5-4B-GGUF |
| UD-Q6_K_XL | 3.9 GBfile size | 4.6 GB3.9 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:UD-Q6_K_XLOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:UD-Q6_K_XLDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-UD-Q6_K_XL.gguf" --local-dir Qwen3.5-4B-GGUF |
| Q8_0 | 4.2 GBfile size | 4.9 GB4.2 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:Q8_0Ollama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:Q8_0Download hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-Q8_0.gguf" --local-dir Qwen3.5-4B-GGUF |
| UD-Q8_K_XL | 5.5 GBfile size | 6.3 GB5.5 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:UD-Q8_K_XLOllama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:UD-Q8_K_XLDownload hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-UD-Q8_K_XL.gguf" --local-dir Qwen3.5-4B-GGUF |
| BF16 | 7.8 GBfile size | 8.6 GB7.8 + 0.25 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/Qwen3.5-4B-GGUF:BF16Ollama ollama run hf.co/unsloth/Qwen3.5-4B-GGUF:BF16Download hf download unsloth/Qwen3.5-4B-GGUF --include "Qwen3.5-4B-BF16.gguf" --local-dir Qwen3.5-4B-GGUF |
MLX for Apple silicon
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| MLX 4bitmlx-community/Qwen3.5-4B-MLX-4bit | 2.8 GBfile size | 3.6 GB2.8 + 0.25 + 0.50 | Download page |
| MLXnace-ai/Qwen3.5-4B-MLX | 5.7 GBfile size | 6.4 GB5.7 + 0.25 + 0.50 | Download page |
Original weights size from the safetensors header, for vLLM, SGLang, Transformers
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| BF16original weights | 8.7 GBfile size | 9.4 GB8.7 + 0.25 + 0.50 |
hf download Qwen/Qwen3.5-4B --local-dir Qwen3.5-4BOther conversions (12; sizes not read)
- RedHatAI/Qwen3.5-4B-quantized.w4a16 Other, w4a16, 173k downloads
- sanskar003/Qwen3.5-4B-AWQ AWQ, 162k downloads
- cyankiwi/Qwen3.5-4B-AWQ-4bit AWQ, 4bit, 142k downloads
- cosmicproc/Qwen3.5-4B-NVFP4 NVFP4, 84k downloads
- cyankiwi/Qwen3.5-4B-AWQ-BF16-INT8 AWQ, bf16, 70k downloads
- AxionML/Qwen3.5-4B-NVFP4 NVFP4, 57k downloads
- QuantTrio/Qwen3.5-4B-AWQ AWQ, 27k downloads
- RedHatAI/Qwen3.5-4B-FP8-dynamic FP8, 9.3k downloads
- litert-community/Qwen3.5-4B Other, 7.0k downloads
- RedHatAI/Qwen3.5-4B-quantized.w8a8 Other, w8a8, 2.7k downloads
- v381654729/qwen3.5-4b-ninfer Other, 1.7k downloads
- lovedheart/Qwen3.5-4B-FP8 FP8, 1.4k downloads
The Hub tags 100 or more quantised repos as conversions of this model; we read the most downloaded and show the maker's own first.
Run it by API
No provider on Hugging Face Inference Providers serves this model (as of 19 Nov 2026, 13:58 UTC). Other hosts may; we only list what the router reports.
Not built yet: one key for every open model, with BlockRun, and our own tested-quality score per provider. Until then, precision and measured speed are the best guide to a provider here.
Fine-tunes of this model
Repos that declare this model as their base, most downloaded first. Listing is not a recommendation; unreviewed ones have not been checked.
- unsloth/Qwen3.5-4B4.7B209kUnreviewed
- org2ai/Wald-4B4.7B111kUnreviewed
- ccore/Qwen3.5-4B-GLB-3D-Coder4.5B65kUnreviewed
- z-lab/Qwen3.5-4B-DFlash44kUnreviewed
- alibiserikbay/JevK54.2B42kUnreviewed
- h2oai/h2o-lightning-4b4.7B26kUnreviewed
- prism-ml/Ternary-Bonsai-2-4B-mlx-2bit14kUnreviewed
- 2Beyes/Aurum-4B-Uncensored4.7B11kUnreviewed
- Qwen/Qwen-Drive-1.0-4B4.5B7.3kUnreviewed
- internlm/Intern-Decision-4B4.5B5.9kUnreviewed
- nvidia/NV-Reason-CT5.3B4.3kUnreviewed
- Jitheswar/Qwen3.5-4B-Amadeus4.5B3.3kUnreviewed
- microsoft/Fara1.5-4B4.5B2.8kUnreviewed
- microsoft/UserRepGen-4B4.5B2.2kUnreviewed
- torchcast-ai/torchcast-decision-4b5.2B1.7kUnreviewed
- inclusionAI/ZwZ2-4B4.5B1.1kUnreviewed
- nvidia/Qwen3.5-4B-Qwen3.8-27B-Draft4.2B757Unreviewed
- MuXodious/Qwen3.5-4B-SOMPOA-heresy-v24.5B495Unreviewed
- KORMo-VL/KORMo-Web-4B5.2B374Unreviewed
- speakleash/Bielik-One-4B-r14.7B341Unreviewed
Planned for this page Not built yet
- Try it in private chat
- A link to Solana AI's confidential chat for the models it serves. This model is not listed as served there.
- Fine-tune it
- Fine-tuning a few base models from this page and publishing the result here.
- Creators and earnings
- Pages for people who publish fine-tunes and adapters of this model, and what they earn.
- Fund and bounties
- Bounties for new open models, paid from a public fund. How the fund will work.
- Contribute compute
- How to give compute to open model work, with Akash.
Model data from the Hugging Face Hub 19 Nov 2026, 13:43 UTC. The repo. Found a mistake on this page? Tell us.