# MiniCPM5-2B by OpenBMB: Open-Weight Model Source: https://savrn.com/models/minicpm5-2b Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve MiniCPM5-2B (2.5B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 5.0 GB | 6.0 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 2.5 GB | 3.0 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 1.3 GB | 1.5 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 7, 2026. [MiniCPM5-2B on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/minicpm5-2b/gpus) ## Model Card By OpenBMB, published under apache-2.0, revision f97400052a43. [MiniCPM Tech Report](https://arxiv.org/pdf/2506.07900) | [MiniCPM Wiki(Chinese)](https://modelbest.feishu.cn/wiki/UtWxwcERfiRIpIkBOjuc3h9tn1D) | [GitHub Repo](https://github.com/OpenBMB/MiniCPM) | [UltraData](https://ultradata.openbmb.cn/) | [Online Demo](https://huggingface.co/spaces/openbmb/MiniCPM5-2B-Demo) English | [中文](https://huggingface.co/openbmb/MiniCPM5-2B/blob/main/README-cn.md) ### Highlights We are releasing MiniCPM5-2B, the second model in the MiniCPM5 series, following [MiniCPM5-1B](https://huggingface.co/openbmb/MiniCPM5-1B). It is a dense 2B Transformer that scales up the same training recipe, built for on-device, local deployment, and resource-constrained scenarios, reaching 2B-class open-source SOTA. 2B-class open-source SOTA: compared with strong open-source models of similar size, MiniCPM5-2B achieves SOTA performance within this comparison set. It remains competitive with 4B-class models overall, while showing its advantages over models of comparable size in coding, mathematics, long-context understanding, tool use, and agentic tasks. [Read the full model card (2,680 words)](https://savrn.com/models/minicpm5-2b/card) ## Configuration Architecture LlamaForCausalLM Context length (tokens) 131,072 Layers 42 Hidden size 2,048 Feed-forward size 6,144 Attention heads 16 Key/value heads 2 Head dimension 128 Vocabulary size 130,560 RoPE base 5,000,000 Stored precision bfloat16 Model type llama ## Identity and Version Repository openbmb/MiniCPM5-2B Publisher OpenBMB Task Text generation Modality Text Library transformers Parameters 2.5B parameters Languages en, zh Revision f97400052a43d642bbc6e9975e2397e3ae6a6b52 First published 2026-09-06 Last updated 2026-09-29 ## Files and Weights 11 files, 5.0 GB in total. The weights are 1 file totalling 5.0 GB in safetensors. Weights1 file · 5.0 GB Configuration4 files · 32.8 KB Tokenizer2 files · 10.0 MB Documentation2 files · 222.5 KB Other1 file · 5.4 KB Repository1 file · 1.5 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model-00000-of-00001.safetensors | Weights | 5.0 GB | 14fb8e7f0a18 | | config.json | Configuration | 704 B | — | | generation_config.json | Configuration | 213 B | — | | model.safetensors.index.json | Configuration | 31.4 KB | — | | special_tokens_map.json | Configuration | 551 B | — | | README-cn.md | Documentation | 110.3 KB | — | | README.md | Documentation | 112.2 KB | — | | chat_template.jinja | Other | 5.4 KB | — | | .gitattributes | Repository | 1.5 KB | — | | tokenizer.json | Tokenizer | 9.9 MB | — | | tokenizer_config.json | Tokenizer | 94.4 KB | — | ## License and Download License apache-2.0 Access Open weights, no gate Download size 5.0 GB [Download from OpenBMB](https://huggingface.co/openbmb/MiniCPM5-2B) Released by OpenBMB through its official repository on Hugging Face. [Read the license](https://www.apache.org/licenses/LICENSE-2.0). ## Built From - Described by arXiv:2506.07900 - Described by arXiv:2602.09003 - Trained on (disclosed) [openbmb/Ultra-FineWeb](https://savrn.com/datasets/ultra-fineweb) - Trained on (disclosed) openbmb/Ultra-FineWeb-L3 - Trained on (disclosed) openbmb/UltraData-Code - Trained on (disclosed) openbmb/UltraData-Math - Trained on (disclosed) openbmb/UltraData-RL-2609 - Trained on (disclosed) openbmb/UltraData-SFT-2605 - Trained on (disclosed) openbmb/UltraData-SFT-Agent-2609 - Trained on (disclosed) openbmb/UltraX-Preview ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 5.0 GB | | 16-bit | 5.0 GB | | 8-bit | 2.5 GB | | 4-bit | 1.3 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Built on This Model - Adapter of[BlazerApex-2B](https://savrn.com/models/blazerapex-2b) - Derived from[BlazerApex-2B](https://savrn.com/models/blazerapex-2b) - Derived from[LycheeAI-coder-2b-II-pro-MLX-bf16](https://savrn.com/models/lycheeai-coder-2b-ii-pro-mlx-bf16) - Derived from[CAT-UT](https://savrn.com/models/cat-ut) - Quantized from[RK182X-LLM-MiniCPM5-2B](https://savrn.com/models/rk182x-llm-minicpm5-2b) - Derived from[RK182X-LLM-MiniCPM5-2B](https://savrn.com/models/rk182x-llm-minicpm5-2b) - Adapter of[MiniCPM5-2B-pjev-LoRA](https://savrn.com/models/minicpm5-2b-pjev-lora) - Derived from[MiniCPM5-2B-pjev-LoRA](https://savrn.com/models/minicpm5-2b-pjev-lora) ## Questions About MiniCPM5-2B ### How much GPU memory does MiniCPM5-2B need? About 6 GB at 16-bit and 1.5 GB at 4-bit: the weights (2.5B parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run MiniCPM5-2B on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### Can I use MiniCPM5-2B commercially? Yes. MiniCPM5-2B is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors. ### What is MiniCPM5-2B's context length? 131,072 tokens, from the maximum position embeddings in its published configuration. ## Similar Models Model · Text generation ### [LFM2.5-2.6B-CyberSec](https://savrn.com/models/lfm2-5-2-6b-cybersec) [Convergent Intelligence](https://savrn.com/model-publishers/reaperdoesntknow) An English LFM2.5 2.6B checkpoint associated with the Trendyol Cybersecurity Instruction Tuning Dataset. This repository contains both Transformers-format model files and local GGUF exports. This is a research release. The repository does not currently publish benchmark, baseline-comparison, or safety-evaluation results. Transformers assets include model.safetensors, configuration files, tokenizer files, and a chat template. Keeping both formats in one repository is convenient, but users should explicitly choose the path that matches their runtime. The checked configuration includes a bitsandbytes quantization block. Pin and test the exact Transformers, Accelerate, bitsandbytes, and device… Open weights apache-2.0 2.7B parameters 131,072 tokens transformers [View model](https://savrn.com/models/lfm2-5-2-6b-cybersec) Model · Text generation ### [Qwen3.5-2B-Opus-Distil](https://savrn.com/models/qwen3-5-2b-opus-distil) [Convergent Intelligence](https://savrn.com/model-publishers/reaperdoesntknow) A 2B-parameter Qwen3.5 fine-tune, part of the Opus-Distil line in the reaperdoesntknow open-weight portfolio. Text-generation / reasoning model, trained with Unsloth + Hugging Face TRL. Open weights apache-2.0 2.3B parameters 262,144 tokens transformers [View model](https://savrn.com/models/qwen3-5-2b-opus-distil) Model · Text generation ### [Qwen3.5-2B-CyberSec](https://savrn.com/models/qwen3-5-2b-cybersec) [Convergent Intelligence](https://savrn.com/model-publishers/reaperdoesntknow) An English Qwen3.5 2B checkpoint associated with the Trendyol Cybersecurity Instruction Tuning Dataset and exported in Transformers / Safetensors format. This release is intended for research and local experimentation. The repository does not currently publish benchmark or safety-evaluation results, so the model should not be treated as a validated cybersecurity authority. The configuration identifies a Qwen3.5 conditional-generation architecture with text and vision components. Use a recent Transformers release that supports this architecture. Dependency and device behavior can vary across Transformers versions. Pin a tested environment for reproducible use. - Research on small-model… Open weights apache-2.0 2.3B parameters 262,144 tokens transformers [View model](https://savrn.com/models/qwen3-5-2b-cybersec) Model · Text generation ### [knivesysl-typed-2b](https://savrn.com/models/knivesysl-typed-2b) [SRSWTI Inc.](https://savrn.com/model-publishers/srswti) local, typed decisions from qwen3.5-2b. one shared state is prefetched once, each question is isolated, every allowed answer is scored as a complete token sequence, and python returns validated choice, score, and noul results. this is an inference system, not rlcd training and not a clone of typesafe jev. it never calls typesafe. the published qwen checkpoint is unchanged; fp8 changes execution precision only. probabilities are normalized support over the candidates you provide, not calibrated correctness probabilities. unlike ordinary autoregressive json generation, the model does not write a response token by token. it scores only the values supplied by the caller. complete-sequence… Open weights apache-2.0 2.3B parameters 262,144 tokens transformers [View model](https://savrn.com/models/knivesysl-typed-2b) Model · Text generation ### [Kimi-K3-DSpark](https://savrn.com/models/kimi-k3-dspark) [RadixArk](https://savrn.com/model-publishers/radixark) A long-context DSpark speculator for Kimi K3. It supports context lengths of up to 1 million tokens. A DSpark speculator for the Kimi K3 target, enabling faster inference through speculative decoding. DSpark extends the DFlash parallel-draft backbone with a Markov logit-bias head and a per-position confidence head. This checkpoint was trained with SpecForge using hidden states from a live SGLang target engine. 64 query heads / 16 KV heads, and blocksize=7 acclen is SGLang's histogram-native request acceptance length, averaged within each question and then equally across questions. RULER V2 uses the 1M input configuration. Actual prompts span 1,000,432–1,047,925 tokens; partition acclen is… Open weights 2.2B parameters 1,048,576 tokens transformers [View model](https://savrn.com/models/kimi-k3-dspark) Model · Text generation ### [ThinkLess-2B](https://savrn.com/models/thinkless-2b) [Sadikh Shaik](https://savrn.com/model-publishers/shaik1903) Qwen3.5-2B that thinks less and answers better. ThinkLess-2B uses 57–82% fewer reasoning tokens than the base model on math and science benchmarks while being more accurate on GSM8K, MATH-500 and GPQA-Diamond, with almost no answers cut off mid-thought. Served with vLLM and its built-in MTP speculative decoding, a typical request finishes 3.4× faster than the base model. It was post-trained in two stages: SFT on the model's own shortest correct solutions (self-distillation, with a same-family 9B teacher filling in the hardest problems), then a short GRPO run with an accuracy-and-length reward. All accuracies are at Qwen's recommended 81,920-token thinking budget; see Evaluation. Mean… Open weights apache-2.0 2.2B parameters 262,144 tokens transformers [View model](https://savrn.com/models/thinkless-2b) ## OpenBMB [All models and datasets](https://savrn.com/model-publishers/openbmb) ## Versions - [f97400052a43](https://savrn.com/models/minicpm5-2b/versions/f97400052a43) · current 2026-10-07 ## Explore More - [All text generation models](https://savrn.com/models/tasks/text-generation) - [All models under apache-2.0](https://savrn.com/models/licenses/apache-2-0) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-10-04. - [Hugging Face record](https://huggingface.co/openbmb/MiniCPM5-2B) - [How the hub is built](https://savrn.com/model-hub/methodology)