MiniCPM5-1B

1.08B parameters, text in, text out, 128K tokens of context.

Verified 2026-10-25 Rank #410 in LLMs · score 79.2 method v1 Sources 3 Downloads 30d 332.7k · #208 of 619 tracked

Changed this week: 3 new quantized builds on the Hub (102 in all).2026-10-24 · 2 new quantized builds on the Hub (101 in all).2026-10-24

Can I run it?0.6GB at INT4Estimatefits a 16 GB cardmethod

Can I use it?Apache 2.0Permissive; patent grantlicence file2026-10-24

Do people use it?332.7kdownloads, 30 days#208 of 619 we track · top 34%+7 likes, downloads −7% this week1,174 likesHugging Face2026-10-24

Worth reading

Picked by rule from 2 tracked pieces: the lab's own post, the most shared, a field report, the latest.

Latest · 4 months ago

Desk-Pet – a local-first desktop pet powered by MiniCPM5

MiniCPM Desk Pet runs the MiniCPM5-1B-GGUF model locally on macOS Apple Silicon or Windows x64 systems requiring about 2 GB disk space.

The app scans the machine for installed coding agents and prompts connection in one click.
Response speed depends on the user's chip, memory pressure, and the selected model.

github.com 2026-07-07 2 HN pts 719 words about the MiniCPM5 family family

2 best of 2 tracked

The 1 pick above, plus these 1. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.

ArticleMiniCPM5-2B – new leading ≤4B modelartificialanalysis.ai · 2026-09-07

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: 102 quantized builds and 100 Spaces.

As a service0no API provider listed
On own hardware102quantized builds
Built on12260 fine-tunes · 62 adapters · 0 merges
In apps100+Hugging Face Spaces

Top Spaces running it: build-small-hackathon/jawbreaker ♥ 87 · openbmb/MiniCPM5-1B-Demo ♥ 38 · build-small-hackathon/tutori ♥ 19 · build-small-hackathon/hackathon-advisor ♥ 16

Will it fit?

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

Fits a 16 GB card at INT4

About 0.6 GB at INT4, 1.3 GB at FP8, 2.6 GB at BF16: weights only, so leave room for the KV cache.

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 30 repo commits so far

Teal is the weights (2026-05-21), amber triangles are 1 dated article (each links to it), grey ticks are repo commits.

What people built on it

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

224 derivatives on the Hub, led by 102 quantized

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

GGUF Abiray/MiniCPM5-1B-GGUF for llama.cpp, Ollama, LM Studio 24.4k downloads MLX mlx-community/MiniCPM5-1B-OptiQ-4bit for Apple silicon 14.5k downloads AWQ cyankiwi/MiniCPM5-1B-AWQ-INT4 for vLLM, SGLang 3.2k downloads

All variants →

Reference

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

What the card says

2 cited sentences

Summary generated, cited

MiniCPM5-1B is a 1.08B parameter LlamaForCausalLM designed for on-device, local deployment and resource-constrained scenarios.^[1]^[2]^[3] The model supports English and Chinese languages and features a context length of 131,072.^[4]^[5]

parameters human
“LlamaForCausalLM”
“on-device, local deployment, and resource-constrained sce…”
languages
“Context Length: 131,072”

Generated 2026-10-07 15:09 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.

Use-case evidence

6 use cases

What it is good at 6 use cases

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

22 fields

Listed in LLMs

Facts

MiniCPM5-1B has 1.08B parameters. (safetensors metadata, observed 2026-10-24)
Context length is 131,072 tokens (128K). (config.json max_position_embeddings, observed 2026-10-24)
License: Apache 2.0 (friction: none). Permissive; patent grant. (license file, observed 2026-10-24)
Created 2026-05-21, last modified 2026-08-17. (HF API, observed 2026-10-24)

Pipeline tag	text-generation
Library	transformers
Architectures	LlamaForCausalLM
Model type	llama
Dtype	bfloat16
Params by dtype	BF16: 1,080,632,832
Size band	S
MoE	no
Context length	131,072 (config.json max_position_embeddings)
Languages	en, zh
Papers	arXiv:2506.07900, arXiv:2602.09003, arXiv:2512.16649, arXiv:2604.13016
License	Apache 2.0 (apache-2.0)
Hub snapshot	2026-10-25

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 MiniCPM5-1B: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)	2.6 GB · Edge (<4 GB)
FP8 (1 byte/param)	1.3 GB · Edge (<4 GB)
INT4 (0.5 bytes/param)	0.6 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: openbmb/MiniCPM5 224 derivatives on the Hub

quantized: 102finetune: 60adapter: 62

Top variants (40 shown, ranked by downloads)

Variant	Relation	Quant	Params	Band	License	Author	Likes	Downloads
Abiray/MiniCPM5-1B-GGUF	quantized	GGUF	—	—	apache-2.0	Abiray	46	24.4k
GnLOLot/MiniCPM5-1B-Claude-Opus-Fable5-Thinking	finetune	—	1.08B	S	apache-2.0	GnLOLot	185	10.7k
mlx-community/MiniCPM5-1B-OptiQ-4bit	quantized	-4bit	1.08B	S	apache-2.0	mlx-community	11	14.5k
GnLOLot/MiniCPM5-1B-Claude-Opus-Fable5-V2-Thinking	finetune	—	1.08B	S	apache-2.0	GnLOLot	47	11.1k
Tdamre/MiniCPM5-1B-litert-lm	finetune	—	—	—	apache-2.0	Tdamre	3	11.3k
hudsongouge/minicpm5-1B-GLM-5.2-Agentic-v8	quantized	GGUF	1.08B	S	other	hudsongouge	3	10.8k
NANI-Nithin/MiniCPM5-1B-GGUF	quantized	GGUF	—	—	—	NANI-Nithin	2	10.5k
tiny-random/minicpm5	finetune	—	4.22M	S	—	tiny-random	2	9.9k
hudsongouge/minicpm5-1B-GLM-5.2-Agentic-v9	quantized	GGUF	1.08B	S	other	hudsongouge	4	8.6k
RemySkye/MiniCPM5-1B-i1-GGUF	quantized	GGUF	—	—	apache-2.0	RemySkye	2	8.7k
saidutta69/MiniCPM5-1B-heretic	quantized	GGUF	1.08B	S	apache-2.0	saidutta69	3	7.8k
ewin-reg/MiniCPM5-1B-Agentic-Tooluse-v3-GGUF	quantized	GGUF	—	—	apache-2.0	ewin-reg	23	6.5k
Luminia/MiniCPM5-1B-Agent-GGUF	quantized	GGUF	—	—	apache-2.0	Luminia	17	6.7k
mradermacher/MiniCPM5-1B-i1-GGUF	quantized	GGUF	—	—	apache-2.0	mradermacher	6	5.6k
mlx-community/MiniCPM5-1B-8bit	quantized	-8bit	1.08B	S	apache-2.0	mlx-community	3	5.5k
hudsongouge/minicpm5-1B-GLM-5.2-Agentic-v7	quantized	GGUF	1.08B	S	other	hudsongouge	1	5.2k
ewin-reg/MiniCPM5-1B-Agentic-Tooluse-Merged-FP16	finetune	-FP16	1.08B	S	apache-2.0	ewin-reg	19	3.6k
hudsongouge/minicpm5-1B-GLM-5.2-Agentic-v1	quantized	GGUF	1.08B	S	other	hudsongouge	0	4.3k
build-small-hackathon/jawbreaker-minicpm5-1b-lora-v8	adapter	—	—	—	mit	build-small-hackathon	7	3.6k
hudsongouge/minicpm5-1B-GLM-5.2-Agentic-v3	quantized	GGUF	1.08B	S	other	hudsongouge	0	3.8k
ewin-reg/minicpm5-gguf-stock-analyst	adapter	gguf	—	—	apache-2.0	ewin-reg	1	3.3k
sahilchachra/MiniCPM5-1B-Uncensored	finetune	—	1.08B	S	other	sahilchachra	4	3.1k
cyankiwi/MiniCPM5-1B-AWQ-INT4	quantized	AWQ	1.08B	S	apache-2.0	cyankiwi	0	3.2k
saidutta69/RaceBench-MiniCPM5	quantized	GGUF	1.08B	S	apache-2.0	saidutta69	0	2.5k
ewin-reg/MiniCPM5-1B-Agentic-DPO-LoRA	adapter	GGUF	1.08B	S	apache-2.0	ewin-reg	0	2.5k
mradermacher/MiniCPM5-1B-GGUF	quantized	GGUF	—	—	apache-2.0	mradermacher	0	2.5k
mlx-community/MiniCPM5-1B-4bit	quantized	-4bit	1.08B	S	apache-2.0	mlx-community	1	2.3k
huihui-ai/Huihui-MiniCPM5-1B-abliterated	finetune	—	1.08B	S	apache-2.0	huihui-ai	10	1.7k
mlx-community/MiniCPM5-1B-mlx-6Bit	quantized	mlx	1.08B	S	apache-2.0	mlx-community	0	2.2k
thealper2/MiniCPM5-1B-Turkish	quantized	GGUF	1.08B	S	apache-2.0	thealper2	3	1.9k
mlboydaisuke/MiniCPM5-1B-CoreAI	quantized	—	—	—	apache-2.0	mlboydaisuke	4	1.8k
K0D3IN/MiniCPM5-1B-heretic	finetune	—	1.08B	S	apache-2.0	K0D3IN	5	1.6k
solomoniw/CallForge-1B-v1	adapter	GGUF	1.08B	S	apache-2.0	solomoniw	0	1.7k
EntityDeletr/MiniCPM5-1B-GGUF	quantized	GGUF	1.08B	S	apache-2.0	EntityDeletr	0	1.7k
nassimjp/MiniCPM5-1B-Pashto	finetune	—	1.08B	S	apache-2.0	nassimjp	1	1.6k
eaddario/MiniCPM5-1B-GGUF	quantized	GGUF	—	—	apache-2.0	eaddario	0	1.4k
hudsongouge/minicpm5-1B-GLM-5.2-Agentic-v5	finetune	—	1.08B	S	other	hudsongouge	1	1.3k
eulogik/yantra-1b-agent	quantized	GGUF	—	—	mit	eulogik	6	1k
koshuro/Minicpm5-1B-Tool-Use	quantized	GGUF	—	—	—	koshuro	1	1.3k
droplychee/droplychee-flash-lite-ultra-for-english	finetune	—	1.08B	S	apache-2.0	droplychee	0	1.3k

Repository changelog

30 commits

changed in the last 7 daysOther: 2026-10-24, 2026-10-24all changes

Repository changelog 30 commits

Date	Commit	SHA
2026-08-17	Update README-cn.md	87179e5
2026-08-17	Update README-cn.md	010676f
2026-08-17	Update README-cn.md	59a3199
2026-08-17	Update README-cn.md	8762e95
2026-08-17	Update README.md	c1838b5
2026-05-26	Update README-cn.md	4e9de7a
2026-05-26	Update README.md	0592600
2026-05-26	fix: update Transformers quickstart to be compatible with transformers v5.x	de0bfbe
2026-05-26	Update README.md	ae28fc1
2026-05-26	fix: update Transformers quickstart to be compatible with transformers v5.x	ce40a6e
2026-05-25	update README	8895d92
2026-05-25	update README	ec01347
2026-05-25	update README	81f7f25
2026-05-25	minor update	b282ae7
2026-05-25	minor update	ab959d9
2026-05-24	Update Chinese model card (README-cn.md)	e23a05a
2026-05-24	Update model card README	b64d06c
2026-05-24	Delete .ms_upload_cache	b265d87
2026-05-24	Update Chinese model card (README-cn.md)	b0834b9
2026-05-24	Update model card README	2a83ad0
2026-05-24	Update Chinese model card (README-cn.md)	243554f
2026-05-24	Update model card README	72ea395
2026-05-24	Update Chinese model card (README-cn.md)	f00e004
2026-05-24	Update model card README	88d0cba
2026-05-24	Update Chinese model card (README-cn.md)	c9a4379
2026-05-24	Update model card README	8773529
2026-05-24	Add Chinese model card (README-cn.md)	ce43f1c
2026-05-24	Add model card README	9da0a79
2026-05-24	Add model card README	94c7e15
2026-05-24	Fix config _name_or_path to openbmb/MiniCPM5-1B	5155e74

Articles and coverage

2 pieces

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.

Analysis1

Desk-Pet – a local-first desktop pet powered by MiniCPM5

github.com2026-07-07authorityabout the MiniCPM5 familyHN 2 pts · 0 comments719 words

MiniCPM Desk Pet runs the MiniCPM5-1B-GGUF model locally on macOS Apple Silicon or Windows x64 systems requiring about 2 GB disk space. The application is distributed under the GNU AGPL-3.0-only license and supports downloading models from Hugging Face or ModelScope during the initial setup. It provides agent-aware reactions to coding tools and includes a neko-style persona adapter for desktop interaction.written, reviewed, cited

The app scans the machine for installed coding agents and prompts connection in one click.
Response speed depends on the user's chip, memory pressure, and the selected model.
Coding-agent reactions depend on each tool's integration behavior and may vary by version.

All coverage tracked for this model (2) — append-only, nothing is ever removed

Title	Source	Published	First seen	Signals
MiniCPM5-2B – new leading ≤4B model	artificialanalysis.ai	2026-09-07	2026-10-07	hn
Desk-Pet – a local-first desktop pet powered by MiniCPM5	github.com	2026-07-07	2026-10-07	hn

Papers

4 papers

Papers 4

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).

MiniCPM4: Ultra-Efficient LLMs on End Devices this model

MiniCPM Team, Chaojun Xiao, Yuxuan Li +80 · 2025-06-09 · PDF · HF paper page (100 upvotes)
Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe cited in card

Yaxuan Li, Yuxin Zuo, Bingxiang He +8 · 2026-04-14 · PDF · HF paper page (116 upvotes)
JustRL: Scaling a 1.5B LLM with a Simple RL Recipe cited in card

Bingxiang He, Zekai Qu, Zeyuan Liu +9 · 2025-12-18 · PDF · HF paper page (34 upvotes)
Data Science and Technology Towards AGI Part I: Tiered Data Management cited in card

Yudong Wang, Zixuan Fu, Hengyu Zhao +14 · 2026-02-09 · PDF · HF paper page (12 upvotes)

Similar models

3 models

Similar models

Llama-3.2-1B-Instruct · 1.24B
same size band (S), same task (text-generation), more providers
gemma-3-1b-it · 1000M
same size band (S), same task (text-generation), more providers
TinyLlama-1.1B-Chat-v1.0 · 1.10B
same size band (S), same task (text-generation), more providers, 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 15:09 UTC · Last verified 2026-10-25

Follow MiniCPM5-1B

Watch this model

/feeds/models/openbmb--MiniCPM5-1B.xml(RSS 2.0) · every change to MiniCPM5-1B
/feeds/families/openbmb--MiniCPM5.xml(RSS 2.0) · every change across the MiniCPM5 family

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