Up to date
Qwen Apache 2.0 out 18 months ago
4.02B parameters, text in, text out, 32K tokens of context. 1 provider serves it.
Verified 2026-10-25 Rank #180 in LLMs · score 90.4 method v1 Sources 3 Downloads 30d 577.7k · #153 of 619 tracked
Changed this week: 1 new quantized build on the Hub (56 in all).2026-10-24
Picked by rule from 18 tracked pieces: the lab's own post, the most shared, a field report, the latest.
OpenAI released gpt-oss-120b and gpt-oss-20b, their first open-weight models since GPT-2 in 2019.
The 2 picks above, plus these 16. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.
Live hosted endpoints, cheapest first, USD per million tokens. Ties break on output price, then provider name.
| # | Cheapest live row | In /1M | Out /1M | Context | Quant | Seen |
|---|---|---|---|---|---|---|
| 1 | Featherless | — | — | — | — | Oct 25 |
Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.
Served by 1 API provider; heavily run locally (56 quantized builds); a big base for builders (645 fine-tunes, adapters and merges).
Top Spaces running it: dqy08/InfoLens1 ♥ 22 · fuvty/C2C_demo ♥ 3 · nics-efc/C2C_demo ♥ 2 · jaredpalmer/kev-4b-decision-demo ♥ 2
Weights only, an estimate: leave room for the KV cache. Method
About 2.4 GB at INT4, 4.8 GB at FP8, 9.7 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
Weights, every dated article, and repo commits on one rail.
Teal is the weights (2025-04-28), amber triangles are 3 dated articles (each links to it), grey ticks are repo commits.
701 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.
Everything else, folded: the full tables, every source.
Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.
Listed in LLMs
| Pipeline tag | text-generation |
|---|---|
| Library | transformers |
| Architectures | Qwen3ForCausalLM |
| Model type | qwen3 |
| Dtype | bfloat16 |
| Params by dtype | BF16: 4,022,468,096 |
| Size band | M |
| MoE | no |
| Context length | 32,768 (config.json max_position_embeddings) |
| Papers | arXiv:2505.09388 |
| License | Apache 2.0 (apache-2.0) |
| Hub snapshot | 2026-10-25 |
vLLM: supported · Qwen3ForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Qwen3ForCausalLM · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed
vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.
Weights-only memory for Qwen3-4B-Base: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 9.7 GB · Single GPU (<24 GB) |
|---|---|
| FP8 (1 byte/param) | 4.8 GB · Single GPU (<24 GB) |
| INT4 (0.5 bytes/param) | 2.4 GB · Edge (<4 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
finetune: 513quantized: 56adapter: 125merge: 7
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| unsloth/Qwen3-Embedding-4B | finetune | — | 4.02B | M | apache-2.0 | unsloth | 4 | 941.3k |
| unsloth/Qwen3-4B-Base | finetune | — | 4.02B | M | apache-2.0 | unsloth | 7 | 703.2k |
| Qwen/Qwen3-Embedding-4B-GGUF | quantized | GGUF | — | — | apache-2.0 | Qwen | 143 | 472.3k |
| unsloth/Qwen3-4B-Base-unsloth-bnb-4bit | quantized | -bnb | 4.12B | M | apache-2.0 | unsloth | 4 | 260.5k |
| tomaarsen/Qwen3-Reranker-4B-seq-cls | finetune | — | 4.02B | M | apache-2.0 | tomaarsen | 11 | 78k |
| xuefengli/qwen3_4b_openthoughts_400k | finetune | — | 4.02B | M | apache-2.0 | xuefengli | 0 | 51.4k |
| Mungert/Qwen3-Embedding-4B-GGUF | quantized | GGUF | — | — | apache-2.0 | Mungert | 5 | 21.1k |
| unsloth/Qwen3-4B-Base-bnb-4bit | quantized | -bnb | 4.14B | M | apache-2.0 | unsloth | 0 | 18.6k |
| QuantFactory/Qwen3-Reranker-4B-GGUF | quantized | GGUF | — | — | apache-2.0 | QuantFactory | 6 | 16.6k |
| Mungert/Qwen3-Reranker-4B-GGUF | quantized | GGUF | — | — | apache-2.0 | Mungert | 3 | 15.4k |
| mlx-community/Qwen3-Embedding-4B-mxfp8 | finetune | — | 4.02B | M | apache-2.0 | mlx-community | 2 | 11k |
| PrimeIntellect/Qwen3-4B | finetune | — | 4.02B | M | apache-2.0 | PrimeIntellect | 1 | 9.8k |
| Mungert/Qwen3-4B-GGUF | quantized | GGUF | — | — | apache-2.0 | Mungert | 8 | 8.3k |
| thewimo/Qwen3-4B-AWQ | quantized | AWQ | 4.02B | M | apache-2.0 | thewimo | 1 | 8.5k |
| AlphaGaO/Qwen3-4B-GPTQ | quantized | GPTQ | 4.02B | M | apache-2.0 | AlphaGaO | 0 | 8k |
| mradermacher/Qwen3-4B-Base-i1-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 2 | 7.9k |
| mozilla-ai/Qwen3-4B-llamafile | finetune | — | — | — | apache-2.0 | mozilla-ai | 12 | 6.8k |
| konantech/Konan-LLM-OND | finetune | — | 4.08B | M | apache-2.0 | konantech | 16 | 6.4k |
| winglian/qwen3-4b-math | finetune | — | 4.02B | M | apache-2.0 | winglian | 0 | 7.1k |
| mlx-community/Qwen3-Reranker-4B-mxfp8 | finetune | — | 4.02B | M | apache-2.0 | mlx-community | 2 | 6.3k |
| YixuanEvenXu/Qwen3-4B-Base-HIP-adapter | adapter | — | — | — | apache-2.0 | YixuanEvenXu | 1 | 6.2k |
| QuantFactory/Qwen3-4B-GGUF | quantized | GGUF | — | — | apache-2.0 | QuantFactory | 3 | 5.8k |
| mradermacher/Qwen3-4B-Base-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 2 | 5.8k |
| dengcao/Qwen3-Reranker-4B-GGUF | quantized | GGUF | — | — | apache-2.0 | dengcao | 0 | 5k |
| HealthDataHub/PARTAGES-Qwen3-4B-PDAPT-SLERP | finetune | — | 4.41B | M | apache-2.0 | HealthDataHub | 0 | 4.6k |
| FastFlowLM/Qwen3-4B-NPU2 | finetune | — | — | — | apache-2.0 | FastFlowLM | 1 | 4.5k |
| SeaFill2025/Qwen3-4B-SFT | finetune | — | 4.41B | M | apache-2.0 | SeaFill2025 | 3 | 4.4k |
| Polygl0t/Tucano2-qwen-3.7B-Base | finetune | — | 3.76B | M | apache-2.0 | Polygl0t | 1 | 4.5k |
| hardlyworking/Sugma4B | finetune | — | 4.02B | M | apache-2.0 | hardlyworking | 0 | 4.6k |
| Thinking-Space/Qwen3-4B-Base-GRPO | finetune | — | 4.41B | M | other | Thinking-Space | 3 | 3.9k |
| budecosystem/hex-1 | finetune | — | 4.02B | M | apache-2.0 | budecosystem | 3 | 3.5k |
| Hastagaras/Qibil-4B-v0.1-RP | finetune | — | 4.02B | M | apache-2.0 | Hastagaras | 0 | 3.6k |
| McGill-NLP/AfriqueQwen-4B | finetune | — | 4.02B | M | cc-by-4.0 | McGill-NLP | 0 | 3.5k |
| chartreuse-verte/prose-rewriter-4b-v1.2 | quantized | GGUF | 4.41B | M | agpl-3.0 | chartreuse-verte | 2 | 3.2k |
| edbeeching/Qwen3-4B-Base-SFT-tr5 | finetune | — | 4.02B | M | — | edbeeching | 0 | 3.2k |
| michaelfeil/Qwen3-Embedding-4B-auto | finetune | — | 4.02B | M | apache-2.0 | michaelfeil | 0 | 3k |
| cmu-lti/osim-4b-mid | finetune | — | 4.41B | M | mit | cmu-lti | 0 | 2.9k |
| MagicCaster/crimeradar-event-merge-qwen3-4b-reasoning-20260629 | finetune | — | 4.02B | M | apache-2.0 | MagicCaster | 0 | 2.8k |
| winglian/qwen3-4b-math-kd-jsd-temp1-v2 | finetune | — | 4.02B | M | apache-2.0 | winglian | 0 | 2.5k |
| dnotitia/Qwen3-4B | finetune | — | 4.02B | M | apache-2.0 | dnotitia | 0 | 2.5k |
| Date | Commit | SHA |
|---|---|---|
| 2025-07-26 | Create LICENSE | 906bfd4 |
| 2025-05-21 | Update README.md | a81b894 |
| 2025-05-20 | update tokenizer_config.json | 0823829 |
| 2025-04-28 | Update README.md | ff84294 |
| 2025-04-28 | Update tokenizer_config.json | 08a151e |
| 2025-04-28 | Upload merges.txt | 467dab3 |
| 2025-04-28 | Update README.md | b1ecb62 |
| 2025-04-28 | Upload folder using huggingface_hub | 2fcc0a0 |
| 2025-04-28 | initial commit | 73e143d |
Hosted endpoints observed for Qwen3-4B-Base, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting.
| Provider | Model ref | Context | Quant | $/M in | $/M out | tok/s | TTFT | Tools | Status | Observed |
|---|---|---|---|---|---|---|---|---|---|---|
| Featherless hf | Qwen/Qwen3-4B-Base | — | — | — | — | — | — | — | live | 2026-10-25 |
Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.
arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).
Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:36 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:03 UTC · Last verified 2026-10-25
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.
