ai:5llm:9slm:9api:5linguaggio:5github:5

Best Small Language Models (SLMs):

📅 2026-07-16 ⏱ 1 min lettura logistar.it

Model Name	Maintainer	Size	Score	VRAM (GB)	Quantized	License	Context Len	Likes	Downloads	Modified	Languages	Architectures
MiniCPM5 1B	openbmb	1B	0.5	2.2	•	apache-2.0	128K	868	354385	2026-05-26	en, zh	LlamaForCausalLM	0
VibeThinker 3B	WeiboAI	3B	0.5	6.2	•	mit	128K	747	63449	2026-06-19	en	Qwen2ForCausalLM	0
HRM Text 1B	sapientinc	1B	0.47	2.4	•	apache-2.0	4K	719	162822	2026-05-21	en	HrmTextForCausalLM	0
Phi 3 Mini 4K Instruct	microsoft	4B	0.46	7.7	•	mit	4K	1436	607430	2025-12-10	en, fr	Phi3ForCausalLM	1081
Hy MT2 1.8B	tencent	2B	0.46	4.1	•	apache-2.0	256K	1131	22767	2026-05-26	zh, en, fr, pt, es, ja, tr, ru, ar, ko, th, it, de, vi,...	HunYuanDenseV1ForCausalLM	0
DeepSeek OCR 2	deepseek-ai	3B	0.45	6.8	•	apache-2.0	8K	1028	3318303	2026-02-03	•	DeepseekOCR2ForCausalLM	0
LocateAnything 3B	nvidia	3B	0.45	7.7	•	other	16K	1967	69443	2026-06-12	en	LocateAnythingForConditionalGeneration	0
Qwen3 ASR 1.7B	Qwen	2B	0.44	4.7	•	apache-2.0	128K	926	1457411	2026-01-30	•	Qwen3ASRForConditionalGeneration	0
SmolLM2 1.7B Instruct	HuggingFaceTB	2B	0.41	3.4	•	apache-2.0	8K	742	154013	2025-04-21	en	LlamaForCausalLM	1062
Ouro 1.4B Thinking	ByteDance	1B	0.41	2.9	•	apache-2.0	64K	45	8015	2026-06-03	•	OuroForCausalLM	1466
MiniMax M2.7 DFlash	nvidia	1B	0.41	2.6	•	other	192K	9	61	2026-07-02	•	DFlashDraftModel	1360
Llama 3.2 3B Instruct	meta-llama	3B	0.4	6.5	•	llama3.2	128K	2324	2169724	2024-10-24	en, de, fr, it, pt, hi, es, th	LlamaForCausalLM	0
Phi 4 Mini Instruct	microsoft	4B	0.39	7.7	•	mit	128K	791	514603	2025-12-10	ar, zh, cs, da, nl, en, fi, fr, de, he, hu, it, ja, ko,...	Phi3ForCausalLM	0
Nanbeige4.1 3B	Nanbeige	3B	0.39	7.9	•	apache-2.0	256K	1142	11138	2026-03-25	en, zh	LlamaForCausalLM	0
LFM2.5 1.2B Instruct	LiquidAI	1B	0.39	2.3	•	other	125K	591	453663	2026-03-30	en, ar, zh, fr, de, ja, ko, es	Lfm2ForCausalLM	0
Qwen2.5 3B Instruct	Qwen	3B	0.38	6.2	•	other	32K	521	6840174	2024-09-25	en	Qwen2ForCausalLM	0
Granite 3.1 2B Instruct	ibm-granite	2B	0.38	5.1	•	apache-2.0	128K	57	11007	2025-04-16	•	GraniteForCausalLM	1178
Gemma 3 1B It	unsloth	1B	0.38	2	•	gemma	32K	22	77185	2025-05-12	en	Gemma3ForCausalLM	1342
HY MT1.5 1.8B	tencent	2B	0.38	4.1	•	•	256K	1184	32205	2026-01-01	zh, en, fr, pt, es, ja, tr, ru, ar, ko, th, it, de, vi,...	HunYuanDenseV1ForCausalLM	0
Gemma3 1B Sft Reasoning 2	Efe2898	1B	0.38	2.6	•	•	32K	•	34	2026-03-26	•	Gemma3ForCausalLM	1432
Qwen2.5 1.5B Instruct	Qwen	2B	0.37	3.1	•	apache-2.0	32K	768	12309706	2024-09-25	en	Qwen2ForCausalLM	0
Llama 3.2 1B Instruct	meta-llama	1B	0.37	2.5	•	llama3.2	128K	1460	8363204	2024-10-24	en, de, fr, it, pt, hi, es, th	LlamaForCausalLM	0

Fonte originale

https://llm-explorer.com/list/?small