Model Follower.

Up to date

Rebuilt
2026-10-25 06:54 UTC
Release radar
117 on watch
Hub snapshot
2026-10-25
Re-checked
daily · every number sourced and dated

Models › SmolLM family

SmolLM-135M

Hugging Face ↗ Follow changes

HuggingFaceTB Apache 2.0 out 2 years ago

135M parameters, text in, text out, 2K tokens of context.

Verified 2026-10-25 Rank #508 in LLMs · score 76.2 method v1 Sources 3 Downloads 30d 79.7k · #341 of 619 tracked

Can I run it?0.1GB at INT4Estimatefits a 16 GB cardmethod
Can I use it?Apache 2.0Permissive; patent grantlicence file2026-10-24
Do people use it?79.7kdownloads, 30 days#341 of 619 we track · below the median+1 likes, downloads −11% this week276 likesHugging Face2026-10-24

Worth reading

Picked by rule from 1 tracked piece: the lab's own post, the most shared, a field report, the latest.

Latest · 2 years ago

SmolLM – Fast and Remarkably Powerful

SmolLM introduces a family of small language models with 135M, 360M, and 1.7B parameters trained on a new high-quality dataset called SmolLM-Corpus.

  • The 1.7B model uses a more traditional architecture while smaller models adopt Grouped-Query Attention and prioritize depth over width.
  • Models trained on curated textbooks outperform web-based textbooks on MMLU and ARC benchmarks.
huggingface.co 2024-07-16 3 HN pts 2.2k words about the SmolLM family family

Where it's used

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.

No API provider serves it, so its downloads are people running it themselves: 23 quantized builds and 37 Spaces.

  • As a service0no API provider listed
  • On own hardware23quantized builds
  • Built on155130 fine-tunes · 24 adapters · 1 merges
  • In apps37Hugging Face Spaces

Top Spaces running it: fromziro/SLM_Regression_Line ♥ 16 · stanley-00/slm-testing ♥ 9 · awacke1/TorchTransformers-CV-SFT ♥ 4 · awacke1/SFT-Tiny-Titans-Agentic-RAG-Party ♥ 3

Will it fit?

Weights only, an estimate: leave room for the KV cache. Method

Fits a 16 GB card at INT4

About 0.1 GB at INT4, 0.2 GB at FP8, 0.3 GB at BF16: weights only, so leave room for the KV cache.

16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 0.1 GB of weights (estimate; KV cache and activations extra)INT40.1 GBFP8: about 0.2 GB of weights (estimate; KV cache and activations extra)FP80.2 GBBF16: about 0.3 GB of weights (estimate; KV cache and activations extra)BF160.3 GB
16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 0.1 GB of weights (estimate; KV cache and activations extra)INT40.1 GBFP8: about 0.2 GB of weights (estimate; KV cache and activations extra)FP80.2 GBBF16: about 0.3 GB of weights (estimate; KV cache and activations extra)BF160.3 GB

Estimate params × bytes per param × 1.2 · how we estimate

The story so far

Weights, every dated article, and repo commits on one rail.

1 dated article and 22 repo commits so far

Teal is the weights (2024-07-14), amber triangles are 1 dated article (each links to it), grey ticks are repo commits.

AugSepOctNovDecJanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2024-08-01: 1 repo commit2024-07-19: 1 repo commit2024-07-16: 16 repo commits2024-07-15: 1 repo commit2024-07-14: 3 repo commitsSmolLM – Fast and Remarkably Powerful · 2024-07-16Weights 2024-07-14 · 2024-07-14Weights 2024-07-14now
AugSepOctNovDecJanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2024-08-01: 1 repo commit2024-07-19: 1 repo commit2024-07-16: 16 repo commits2024-07-15: 1 repo commit2024-07-14: 3 repo commitsSmolLM – Fast and Remarkably Powerful · 2024-07-16Weights 2024-07-14 · 2024-07-14Weights 2024-07-14now

What people built on it

178 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

178 derivatives on the Hub, most of them fine-tunes

From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.

23 quantized (Hugging Face model tree)130 fine-tunes (Hugging Face model tree)1 merge (Hugging Face model tree)24 adapters (Hugging Face model tree)23 quantized130 fine-tunes1 merge24 adapters
23 quantized (Hugging Face model tree)130 fine-tunes (Hugging Face model tree)1 merge (Hugging Face model tree)24 adapters (Hugging Face model tree)23 quantized130 fine-tunes1 merge24 adapters
GGUF nisten/Biggie-SmoLlm-0.15B-Base for llama.cpp, Ollama, LM Studio 63.3k downloads MLX paperbd/smollm_135M_neuraltxt_mlx_v1 for Apple silicon 384 downloads AWQ TechxGenus/SmolLM-135M-Instruct-AWQ for vLLM, SGLang 465 downloads

All variants →

Reference

Everything else, folded: the full tables, every source.

Use-case evidence

1 use case

What it is good at 1 use case

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.

Facts

20 fields

Listed in LLMs

Facts

Pipeline tagtext-generation
Librarytransformers
ArchitecturesLlamaForCausalLM
Model typellama
Dtypebfloat16
Params by dtypeF32: 134,515,008
Size bandS
MoEno
Context length2,048 (config.json max_position_embeddings)
Languagesen
LicenseApache 2.0 (apache-2.0)
Hub snapshot2026-10-25

Servability

vLLM · SGLang

Servability

vLLM: supported · LlamaForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · LlamaForCausalLM · 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.

Hardware fit

3 estimates

Hardware fit ESTIMATE

Weights-only memory for SmolLM-135M: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)0.3 GB · Edge (<4 GB)
FP8 (1 byte/param)0.2 GB · Edge (<4 GB)
INT4 (0.5 bytes/param)0.1 GB · Edge (<4 GB)

Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.

All variants

40 variants

Lineage

Original model Family: HuggingFaceTB/SmolLM 178 derivatives on the Hub

quantized: 23finetune: 130adapter: 24merge: 1

Top variants (40 shown, ranked by downloads)

Relation
Quant
40 rows
Variant Relation Quant Params Band License Author Likes Downloads
HuggingFaceTB/SmolLM-135M-Instruct quantized — 135M S apache-2.0 HuggingFaceTB 147 2.2M
nisten/Biggie-SmoLlm-0.15B-Base quantized GGUF — — mit nisten 241 63.3k
unsloth/SmolLM-135M finetune — 135M S apache-2.0 unsloth 4 36.1k
unsloth/SmolLM-135M-bnb-4bit quantized -bnb 138M S apache-2.0 unsloth 0 26.4k
LemiSt/SmolLM-135M-de finetune — 135M S apache-2.0 LemiSt 1 13k
mradermacher/SmolLM-135M-GGUF quantized GGUF — — apache-2.0 mradermacher 1 11.3k
QuantFactory/Biggie-SmoLlm-0.15B-Base-GGUF quantized GGUF — — — QuantFactory 6 8.2k
Joshfcooper/ai-text-detector-optimized quantized — — — apache-2.0 Joshfcooper 0 6.2k
muqiann/SmolLM-135M-Q4_K_M-GGUF quantized GGUF — — apache-2.0 muqiann 0 4.1k
mradermacher/SmolLM-135M-i1-GGUF quantized GGUF — — apache-2.0 mradermacher 1 3.2k
distily/distily_smollm_dataset_sweep finetune — 81.4M S creativeml-openrail-m distily 0 1.9k
paperbd/smollm_135M_neuraltxt_v1 finetune — 135M S mit paperbd 3 1.7k
bcatt/business-news-generator finetune — 135M S apache-2.0 bcatt 0 1.5k
vxef/smol_lora_toy adapter — — — — vxef 0 1.4k
Dominic/smollm135_fullprec_tinystories finetune — 135M S apache-2.0 Dominic 0 1.2k
Vigor26/business-news-generator finetune — 135M S apache-2.0 Vigor26 0 1.1k
Niansuh/Biggie-SmoLlm-0.15B-Base quantized GGUF — — mit Niansuh 0 1k
sawalni-ai/smollm-fw-darija finetune — 135M S — sawalni-ai 2 916
distily/distily_seq_len_batch_size finetune — 81.4M S creativeml-openrail-m distily 0 990
bcatt/business-news-generator-v1 finetune — 135M S apache-2.0 bcatt 0 910
onnx-community/SmolLM-135M-ONNX quantized — — — — onnx-community 2 784
tvaldez/business-news-generator finetune — 135M S apache-2.0 tvaldez 0 779
farpluto/SmolLM-135M-Instruct-Finetune-LoRA adapter — — — apache-2.0 farpluto 0 764
Corizfuo/SmolLM-135M-Instruct finetune — 135M S apache-2.0 Corizfuo 0 688
sq-ss/business-news-generator finetune — 135M S apache-2.0 sq-ss 0 635
MarioAlviano/business-news-generator finetune — 135M S apache-2.0 MarioAlviano 0 547
TechxGenus/SmolLM-135M-Instruct-AWQ quantized AWQ 163M S apache-2.0 TechxGenus 0 465
genaibook/business-news-generator finetune — 135M S apache-2.0 genaibook 0 457
abarget/business-news-generator finetune — 135M S apache-2.0 abarget 0 432
Thox-ai/ThoxNano-135M finetune GGUF 135M S apache-2.0 Thox-ai 0 432
Siddartha10/epoch_1 finetune — 135M S apache-2.0 Siddartha10 0 431
StarryNightDelfina/business-news-generator finetune — 135M S apache-2.0 StarryNightDelfina 0 419
paperbd/neuraltxt-v1-135M finetune — 135M S mit paperbd 0 419
OpenMOSS-Team/SmolLM-135M-MLA-d_kv_8 finetune — 129M S apache-2.0 OpenMOSS-Team 1 366
manzaldua95/business-news-generator finetune — 135M S apache-2.0 manzaldua95 0 400
mjschock/SmolLM-135M-Q4_K_M-GGUF quantized GGUF — — apache-2.0 mjschock 1 339
jjjjjvvvvv/business-news-generator finetune — 135M S apache-2.0 jjjjjvvvvv 0 384
paperbd/smollm_135M_neuraltxt_mlx_v1 finetune — 135M S mit paperbd 0 384
bjbjbj/business-news-generator finetune — 135M S apache-2.0 bjbjbj 0 370
Noahship/smollm-agnews finetune — 135M S apache-2.0 Noahship 0 358

Repository changelog

22 commits

Repository changelog 22 commits

DateCommitSHA
2024-08-01 Update README.md 1d46172
2024-07-19 Update tokenizer_config.json 0ccfeb9
2024-07-16 Update README.md eec6e46
2024-07-16 Update README.md 209e03a
2024-07-16 Update README.md 164f715
2024-07-16 Update README.md ed79565
2024-07-16 Update README.md 9dfc659
2024-07-16 Update README.md 6e67a6f
2024-07-16 Update README.md 79e48c3
2024-07-16 Update README.md 004e26f
2024-07-16 Update README.md 40722a7
2024-07-16 Update README.md 206152b
2024-07-16 Update README.md 08113ae
2024-07-16 Update README.md 1f21d13
2024-07-16 Update README.md f33ac84
2024-07-16 Update README.md a5d92e1
2024-07-16 Update README.md 2229821
2024-07-16 Update config.json 501d302
2024-07-15 Upload ONNX weights (#1) 1476377
2024-07-14 Upload tokenizer f9d39ea
2024-07-14 Upload LlamaForCausalLM 9180459
2024-07-14 initial commit ae2ee32

Articles and coverage

1 piece

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.

Official1

SmolLM – Fast and Remarkably Powerful
huggingface.co2024-07-16officialauthorityabout the SmolLM familyHN 3 pts · 0 comments2,234 words
SmolLM introduces a family of small language models with 135M, 360M, and 1.7B parameters trained on a new high-quality dataset called SmolLM-Corpus. The 135M and 360M models are each trained on 600B tokens, while the 1.7B model is trained on 1T tokens from the corpus. All three models utilize a tokenizer with a vocab size of 49152 and a context length of 2048 tokens.written, reviewed, cited
  • The 1.7B model uses a more traditional architecture while smaller models adopt Grouped-Query Attention and prioritize depth over width.
  • Models trained on curated textbooks outperform web-based textbooks on MMLU and ARC benchmarks.
  • SmolLM-135M outperforms MobileLM-125M despite being trained on only 600B tokens compared to MobileLM's 1T tokens.

Similar models

3 models

Similar models

  • gpt2 · 137M
    same size band (S), same task (text-generation), permissive license
  • SmolLM2-135M · 135M
    same size band (S), same task (text-generation), permissive license
  • gpt-neo-125m · 150M
    same size band (S), same task (text-generation), permissive license

Sources

3 sources

Sources

Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:39 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:46 UTC · Summary: 2026-10-07 15:10 UTC · Last verified 2026-10-25

Follow SmolLM-135M

Watch this model

Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.