752MQwen
Qwen3-0.6B
2025-04-27 · ♥ 1764 · ⬇ 30,768,858 · apache-2.0
Run with llama.cpp →Want to try a small model right now? WebSLM runs them in your browser with llama.cpp. No install, nothing leaves your device.
Trending small models
Most-trending text models with ≤4B parameters on Hugging Face.
4.1BXHToken
Spark-X2.5-4B
2026-08-24 · ♥ 1396 · ⬇ 28,829 · apache-2.0
Run with llama.cpp →149MAxiomicLabs
GPT-X3-150M
2026-09-23 · ♥ 23 · ⬇ 53 · apache-2.0
Run with llama.cpp →146Mbench-labs
cagliostro-v3.5
2026-10-04 · ♥ 22 · ⬇ 695 · apache-2.0
Run with llama.cpp →2.5Bopenbmb
MiniCPM5-2B
2026-09-06 · ♥ 1732 · ⬇ 1,282,719 · apache-2.0
Run with llama.cpp →3.2Bmeta-llama
Llama-3.2-3B-Instruct
2024-09-18 · ♥ 2762 · ⬇ 1,513,998 · llama3.2
Run with llama.cpp →1.1BIFM
K2-Type-0.9B
2026-09-26 · ♥ 33 · ⬇ 1,814 · apache-2.0
Run with llama.cpp →1.2Bmeta-llama
Llama-3.2-1B-Instruct
2024-09-18 · ♥ 1808 · ⬇ 6,976,807 · llama3.2
Run with llama.cpp →2.7BLiquidAI
LFM2.5-2.6B
2026-07-28 · ♥ 807 · ⬇ 118,821 · other
Run with llama.cpp →137Mopenai-community
gpt2
2022-03-02 · ♥ 4221 · ⬇ 15,521,105 · mit
Run with llama.cpp →146Mbench-labs
cagliostro-v3.5-chat
2026-10-08 · ♥ 12 · ⬇ 427 · apache-2.0
Run with llama.cpp →1.9Bnerkyor
Qwen3.8-27B-EfficientThink-Uncensored-K3-Opus5-Grok4.6-GPT5.6Sol-SFT-SimPO-DFlash2
2026-09-03 · ♥ 59 · ⬇ 30,622 · apache-2.0
Run with llama.cpp →Research
Latest arXiv papers on small and on-device language models.
- PIVOT: Perplexity-Informed KD-to-RL Transition Scheduling for Vertical-Domain Few-Shot Distillation
- Grammar Concept Annotation at Scale: Deployed Fine-Tuned Small Language Models Outperform Prompted Frontier Models
- Skill-SLM: Agent Skill-driven Small Language Models for Reliable Robot Operation
- Using Small Language Models to Reverse-Engineer Machine Learning Pipelines Structures
- Decoupling Logic from Persona: Structural Immunity of Edge LLM Agents to Context Pollution
- Multi-Objective Aligned Small Language Model Framework for SUD Patient Dialogue Generation
- ToolRACER: A Robust Agentic Conversation Emulation Resource for Agent Training and Evaluation
- Multi-Label Topic Assignment via LLM Distillation: A Comparative Analysis of Generative vs. Discriminative Student Models
- How Fragile Is On-Device Language Model Safety? Localizing Safety-Critical Parameters for Sparse Fault Analysis
- A Systematic Study of Small Language Models on Abstract Reasoning Tasks
- Do I Need the Cloud? Uncertainty-Aware Step-Level Handoff for Small Language Model Agents
- Which and When to Admit: Gradient Admission for Data-Centric Small Language Model Finetuning
- Evaluating Escalation Signals for LLM Routing: Targets, Controls, and Five Ways to Fool Yourself
- SAFESHIELD: A Decision-Organization Framework for Deployment-Time Safety of Small Language Models
- Learning What to Imitate: Entropy-Aware Distribution Mixing
Community
Hacker News discussions about small models.
From the blogs
Hugging Face blog posts mentioning small or on-device models.
- Transformers now runs llama.cpp quants
- Granite 4.2 LLMs: How They're Built
- Up to 3.2x Faster Inference with LFM2.5-DSpark
- Hugging Face and Cerebras bring Gemma 4 to real-time voice AI
- Granite Embedding Multilingual R2: Open Apache 2.0 Multilingual Embeddings with 32K Context — Best Sub-100M Retrieval Quality
- Granite 4.1 LLMs: How They’re Built
- Welcome Gemma 4: Frontier multimodal intelligence on device
- Granite 4.0 3B Vision: Compact Multimodal Intelligence for Enterprise Documents