OpenBMB (Open Lab for Big Model Base) aims to build foundation models and systems towards AGI. Connect with us: https://xitter.twitterapi.xyz/t.co/N9pevTnoOa
Joined February 2022
- Tweets2.3K
- Following408
- Followers13.3K
- Likes3.2K
Pinned Tweet
🌍 MiniCPM Ambassador Program S2 is now open for applications!
Are you building with AI models, agents, or on-device AI? Do you enjoy sharing what you learn and helping fellow developers? We’d love to have you on board!
Join the OpenBMB community to build with MiniCPM, share your experience, and inspire more developers to turn ideas into reality. 💡
✨ What’s in it for you?
📣 Greater visibility: Showcase your projects through OpenBMB’s official channels and global developer events.
🤝 Closer connections: Connect directly with the MiniCPM team and developers worldwide.
🛠️ Technical resources: Get deployment assistance and 1,000 MiniCPM API credits for your experiments and demos.
🎪 Community support: Access event kits, promotional support, and funding for community activities.
🏅 Official recognition: Earn an exclusive MiniCPM Ambassador badge and official certificate, becoming a recognized member of the OpenBMB global developer ecosystem.
💡 How can you contribute?
Write tutorials. Build applications. Contribute code. Share feedback. Bring developers together.
Whatever your strengths, your ideas and hands-on experience matter. Let's grow together, empower others, and keep open source thriving. 💙
🔗 Apply now:
docs.google.com/forms/d/1U9P…
#MiniCPM #OpenBMB #OpenSource #AI
📸 Snap your meal. Review the estimate. See your daily intake at a glance.
Our ambassador @emwstudio built esp32-intake, a food-logging demo that connects MiniCPM-V’s visual understanding with an iOS app and a circular ESP32 display. 🍽️
Here’s the workflow:
🧠 MiniCPM-V 4.7 35B-A3B identifies food in a photo and estimates calories through the ModelBest API.
✅ You review the estimate before adding it to your daily calorie total. Re-estimate or skip it, with a preview of how the meal would affect your daily target.
📊 The app stores today’s calorie total locally and sends the total and daily target over the local Wi-Fi network to an ESP32-S3 AMOLED display.
🟢🟡🔴 A progress ring, calories remaining — or over target — and matching status colors make the day’s intake easy to check.
The model runs online via API; the ESP32 handles the display. Calorie estimates are approximate and intended for general reference.
We love seeing our ambassadors turn visual intelligence into hands-on experiences that connect software, hardware, and everyday routines. 💙
🛠️ Explore @emwstudio’s project:
github.com/emwstudio/esp32-i…
🚀 MiniCPM-V GitHub:
github.com/OpenBMB/MiniCPM-V
🤗 Model on Hugging Face:
huggingface.co/openbmb/MiniC…
#MiniCPM #OpenBMB #ESP32 #OpenSource
📄 Can your document agent show why a result should be trusted and flag what needs another look?
Community developer @mucerx built TraceCell, a local-first document parsing and verification agent powered by MiniCPM. 🔍
Three models work together:
📖 MiniCPM-OCR 1.0 parses document content.
⚖️ MiniCPM-OCR Jev scores recognition candidates.
🧠 MiniCPM-V 4.7 reviews ambiguous regions, with its answers scored again before acceptance.
📍 Every field links back to source-image coordinates, confidence, and review status in an evidence ledger.
🔄 Uncertain results trigger targeted re-parsing, model review, or human escalation. Documents are routed for automatic acceptance, spot checks, or full review.
🧮 TraceCell also checks arithmetic and reconciles documents—catching mismatches even when the text itself was recognized correctly.
🧪 The author’s current demo covers 5 synthetic documents, 83 regions, and 52 business fields. Calibration on 78 labelled samples reports 80.77% automatic acceptance with no incorrect acceptances. These are sample-set results, not independently verified production guarantees.
With MCP and Skill support, it’s a practical community example of turning visual understanding into traceable decisions. 💙
🛠️ Explore TraceCell:
github.com/logicore-code/Tra…
🚀 Build with MiniCPM:
GitHub — MiniCPM-OCR:
github.com/OpenBMB/MiniCPM-O…
GitHub — MiniCPM-V:
github.com/OpenBMB/MiniCPM-V
🤗 Hugging Face — MiniCPM-OCR 1.0:
huggingface.co/openbmb/MiniC…
#MiniCPM #OpenBMB #DocumentAI #OpenSource
OpenBMB retweeted
Haven't received a trophy for a long time! MiniCPM ambassador trophy received 🫡
Appreciate the sentiment and the acknowledgement from the @OpenBMB team.
I look forward to building more products that help with my mission to build AI for the good of humanity and the world.
Small models are the way of the future and I'm glad there are labs like OpenBMB who make them a priority.
Love to see what the community builds with these small language models.
Your private documents should stay private. No cloud uploads, no Wi-Fi, no compromise. Just drop a scanned document into @Nativ_AI and let MiniCPM-V 4.7 handle the OCR locally on your Mac. This is exactly the kind of practical, privacy-first AI use case we love to see!
That scanned contract you'd never paste anywhere? Let your Mac read it, Wi-Fi off.
Drop the page into Nativ and @OpenBMB's MiniCPM-V 4.7 turns it into text you can search and edit, locally on your Mac.
runnativ.ai
This is awesome, Luis! 🚀 Love seeing MiniCPM-OCR integrated into Hermes Agent. Thanks for building this and sharing it with the community! 💙
Introducing the MiniCPM-OCR plugin for Hermes Agent.
This is a problem I have experienced myself: LLMs by themselves are not always very good at OCR; sometimes I have needed another model specialized in it to solve complex document parsing.
For example, recently I have had difficulties understanding some physician recipes and other manuscripts.
This is where MiniCPM-OCR comes in. You can use the model and harness you already have to utilize a best-in-class OCR model by @OpenBMB.
@Teknium @NousResearch
Code: github.com/PovedaAqui/minicp…
PR: github.com/NousResearch/herm…
📸 Your photo archive holds years of experience. What if you could ask it questions—and revisit the stories behind them?
Community developer @sgfvdsg86265 built LifeLog, a local-first life-review engine inspired by 29,294 photos collected over 10 years. Powered by MiniCPM, it turns an overwhelming album into an archive you can explore through language. 💡
👀 Understand the moments: MiniCPM-V 4.7 1B generates structured descriptions of scenes, people, visible location clues, activities, and overall atmosphere.
🧩 Discover the connections: Photo descriptions are organized into days and life episodes, helping reconnect scattered memories.
✍️ Tell the stories: MiniCPM-V 4.5 8B adds deeper interpretation, generates three narrative styles, and creates a personal profile supported by photo details.
💬 Ask, then check: LifeLog supports Chinese-language Q&A over photo records, returning answers alongside relevant records. Its demo pairs workplace and travel insights with photos, timestamps, and descriptions for review.
🔒 Keep memories local: Photo understanding and writing run offline on a 16GB Apple Silicon Mac.
Visual understanding, evidence-backed retrieval, and narrative generation—all working together to help people rediscover their own experiences. That’s the kind of everyday intelligence we love seeing our community build. 💙
🛠️ Explore LifeLog:
github.com/lengjingxu/lifelo…
🚀 Build with MiniCPM-V:
github.com/OpenBMB/MiniCPM-V
🤗 Hugging Face:
huggingface.co/openbmb/MiniC…
#MiniCPM #OpenBMB #LocalAI #OpenSource
Excited to see MiniCPM-OCR integrated into Docling! 🚀💙 Thanks for making open-source document processing even more accessible! 🙌
Docling v2.149.0 just shipped and who should update depends on what you feed it 📄
Building RAG over PDFs?
New opt-in --use-embedded-picture-images pulls a figure straight out of the PDF at native pixels instead of a rendered crop. If the region also holds text or vector shapes it falls back to the crop
Need OCR on one consumer GPU?
New minicpm_ocr preset for OpenBMB's 1.3B MiniCPM-OCR (English + Chinese):
docling doc.pdf --pipeline vlm --vlm-model minicpm_ocr
Transcribing audio on a Mac?
ASR can now run on mlx-audio with no torch, and Granite Speech recordings get split into ~30s windows
⚠️ Scripts: --vlm-model X now really switches to the VLM pipeline (it used to be silently ignored), and pairing it with --pipeline standard now errors
pip install -U docling
github.com/docling-project/d…
Excited to see MiniCPM-OCR 1.0 supported in vLLM v0.33.0! 🚀 Huge thanks to the vLLM team and all contributors for making open-source AI more accessible! 💙
vLLM v0.33.0 is out! 🚀
927 commits · 389 contributors · 122 first-timers
🤖 New models: Reflection Beam-1.0, Step-5 Preview, Nemotron Parse 3.5, MiniCPM-OCR 1.0
🔥 PyTorch 2.14 — CUDA 13.0-only wheels, libtorch ABI-stable
🎲 Spec decoding: XPress refiner, FLy lossy verify (1.04–1.31x), DSpark 1373→3390 tok/s on GB300
⚡ NVIDIA decode: 1.5–2x latency decode on GB300, FA4 sparse MLA, 39–48% lower TTFT
🚀 New AWS EFA image + shorter vllm/vllm name
🐧 ROCm 10.0 default, MonoKernel 2.32x P50
Full notes ➡️ github.com/vllm-project/vllm…
🔧 Can an LLM agent debug a broken RL training run?
Introducing AutoRLBench — a unified benchmark putting 10 frontier agents to the test on real-world, failed RL runs.
Full code, configs, logs, and metrics are kept intact. No synthetic bugs. The goal? Pinpoint the exact root cause behind deceptively "healthy-looking" reward curves.
📊 Key findings:
- Even the best frontier model barely reaches ~40%, with a massive capability gap across tiers.
- Models perform best on explicit Config bugs, but struggle severely on implicit Data bugs.
- Stronger agents diagnose faster and consume fewer tokens than weaker ones.
In RL, the hardest failure to catch is the one that still looks like learning.
📄 Website: thinking-space.github.io/Aut…
💻 Code: github.com/Thinking-Space/Au…
#RL #LLM #Agent #Benchmark #PostTraining #RSI
Awesome work, Josh! 💙 Love seeing MiniCPM5-2B and MiniCPM-OCR come together in a fully local, privacy-focused application. Congrats on the v1.0 release! 🚀
materia release 1.0.0: blood panel in, cited supplement and herb options out.
every claim traceable to a source, withheld when the citation doesn't hold.
100% open weights on CPU: @OpenBMB MiniCPM5-2B for chat and reranking, MiniCPM-OCR-1.0 for lab-report OCR, llama.cpp, SQLite BM25, a deterministic safety screen.
Not a diagnosis, not a prescription. Local, no cloud, 100% private.
github.com/RegeneratusLabs/m…
Love this idea! 💙 Giving agents a way to organize their memories offline with MiniCPM is such a creative use case. Excited to see minicpm-dream in Spark S2! 🚀
Next @OpenBMB based project
minicpm-dream.
offline dream memory tool. your agent forgets the day; this gives it back. A nightly pass (based on the existing dreaming memory protocol) it reads the day's conversations, keeps the facts that matter, and writes a diary you can read all on your own machine with a small MiniCPM model.
currently optimised to plug into @NousResearch Hermes - but open to contributions for more harnesses.
offline local models sorting your memories while you sleep.
I will be entering this into the OpenBMB Sprark Program S2 hackathon! Alongside my recent materia project.
follow along for more fun small language model projects
github.com/RegeneratusLabs/m…
Great perspective! 💙 Speed, accuracy, and cost all matter for real-world document processing. Excited to see MiniCPM-OCR integrated into LiteParse. Thanks for bringing it into the ecosystem! 🚀
The world needs two types of document OCR:
1. Extremely fast/cheap OCR tools: Each organization might have trillions of documents, and this is oftentimes helpful as a 'first pass' to digitalize each document so that it's accessible to AI agents. Accuracy might not be the highest, but it's 'good enough' for agentic retrieval and a deeper second pass.
This includes free/OSS tools like pypdf, pymupdf. Over the past year, we've invested heavy efforts in this area through `liteparse` - the fastest/most accurate free/OSS parser. It integrates with 5 other open-weight OCR tools, most recently with MiniCPM-OCR from @OpenBMB
github.com/run-llama/litepar…
We're also doing a lot more in this space - stay tuned!
2. Accurate VLM-based OCR: This includes the subset of documents that are a) complex, and b) the organization must parse accurately. The data needs to be accurate because this data is used in business critical workflows across finance, insurance, healthcare, and more.
At the same time, even if the tool is not free, any VLM needs to be able to scale to billions of pages in a cost-effective manner.
Two weeks ago we launched OCRia-1, the best general purpose model for document parsing. It not only parsers complex text layouts, complex tables, but also visual diagrams and charts.
llamaindex.ai/blog/parse-ocr…
It powers all our modes within LlamaParse; see the most recent benchmarks on ParseBench: parsebench.ai/
-------
Both of these document OCR types require rigorous benchmarks, data+training pipelines, and an ambitious/talent-dense team pushing the pareto frontier of accuracy and cost.
If you're looking to solve document understanding across either frontier, come talk to us!
Two awesome builds! 🔥 Love seeing MiniCPM power such different use cases, all running locally and offline. Thanks for building and sharing! 💙🚀
My @OpenBMB MiniCPM Spark S2 program submissions! 2 different angles - help the health of a human or help the memory of an agent.
all offline, all local, all powered & optimised for the MiniCPM family!
Lets gooo!
Get your submissions in.
github.com/RegeneratusLabs/m…
blood panel in, cited supplement and herb options out. a naturopathic or integrative doctor AI assistant that works completely offline.
every claim traceable to a source, withheld when the citation doesn't hold.
100% open weights on CPU: OpenBMB MiniCPM5-2B for chat and reranking, MiniCPM-OCR-1.0 for lab-report OCR, llama.cpp, SQLite BM25, a deterministic safety screen.
github.com/RegeneratusLabs/m…
Your agent forgets the day; this gives it back. A nightly pass reads the day's conversations, keeps the facts that matter, and writes a diary you can read — all on your own machine with a small MiniCPM model.
Huge thanks to @llama_index for integrating MiniCPM-OCR into LiteParse! Excited to see MiniCPM-OCR enabling fast, high-quality document parsing. Looking forward to seeing more builders put it to use! 🚀
LiteParse integrates with MiniCPM-OCR from @openBMB 🚀
MiniCPM-OCR is a new lightweight OCR model that performs OCR after capturing cropped layout regions. On ParseBench, it beats Gemini 3.5 Flash Lite and PaddleOCR-VL-1.6 on content faithfulness.
LiteParse is our ultra-fast OSS document parser. When OCR is required, LiteParse lets you connect any existing OCR model.
See the integration: github.com/run-llama/litepar…
2
14
1,176
📄 Your documents, your comparison.
Try MiniCPM-OCR 1.0 side by side with other OCR models on these community-built HF Spaces, no setup needed! 🔍
Thanks @yangzhizheng1 for sharing. Bring your scans, tables, and formulas, and let us know what you find! 🙌
New open-source OCR models are dropping fast!
Try MiniCPM-OCR 1.0 (OpenBMB, 1.3B) side-by-side with competitors on these free HF Spaces — no setup needed:
MiniCPM-OCR vs OXR
huggingface.co/spaces/pzc163…
MiniCPM-OCR vs Unlimited-OCR-2
huggingface.co/spaces/nielsr…
Upload real documents screenshot (scans, photos, tables, formulas) and compare parsing quality. Share your results!
#OCR #DocumentAI #MiniCPM #HuggingFace #OXR #UnlimitedOCR2
Excited to see MiniCPM-OCR integrated into LiteParse! 🚀💙 Thanks to the LlamaIndex team for making document parsing even more accessible to developers. 🙌
LiteParse integrates with MiniCPM-OCR from @openBMB 🚀
MiniCPM-OCR is a new lightweight OCR model that performs OCR after capturing cropped layout regions. On ParseBench, it beats Gemini 3.5 Flash Lite and PaddleOCR-VL-1.6 on content faithfulness.
LiteParse is our ultra-fast OSS document parser. When OCR is required, LiteParse lets you connect any existing OCR model.
See the integration: github.com/run-llama/litepar…
🤖 What happens when you hand an agent the full SFT data-research task?
It only mixes ratios. No repaired samples, no new responses, no synthetic data — a word cloud full of "weights" and "proportions", and almost no score gain over dozens of rounds.
AutoSFT (THUNLP × BUPT × ModelBest × UIUC × CUHK) splits the data-research lifecycle into roles: a dedicated Executor, 4 Analysis Subagents (training data / failure cases / learning dynamics / research history), a Synthesis writer with a verifier, and a meta agent that picks the next recipe.
1️⃣ Single-agent baseline: 33 rounds, only Mixture; context hit 94.7%, compressed 10 times; Executor cuts compression 10→5 and shell events 58%→29%.
2️⃣ AutoSFT: all 6 action families appear across 30 rounds; IFEval round 6 = 63.33, beating the baseline's best over 50 rounds; +6.5 at the same budget; round 30 = 67.10.
3️⃣ Synthesis alone: IFEval +3.4, IFBench +2.9; MBPP+ coding +3.7.
The takeaway: automating data research may depend more on covering diverse data actions than on a stronger single action.
📄 Paper: arxiv.org/abs/2610.22194
💻 Code: github.com/Thinking-Space/Au…
#AIAgent #SFT #LLM #DataCentricAI #PostTraining #RSI #AgenticAI #THUNLP #AutoSFT #FineTuning
🐱 Who’s been drinking from the water bowl? MiniCPM-V helps keep track.
Our ambassador @emwstudio built esp32-eye, an open-source monitor that recognizes individual cats and logs their daily drinking events, using MiniCPM-V 4.7, two ESP32-S3 boards, and a MacBook. 💧
From camera feed to daily counts, here’s how it works:
📷 Capture: An ESP32-S3 camera streams video from beside the water bowl to the MacBook.
⚡ Detect: MiniCPM-V 4.7 1B runs locally on the MacBook to check whether a cat is in view, filtering frames before calling the larger model.
🔍 Identify: MiniCPM-V 4.7 35B-A3B checks the scene and compares reference photos to identify the cat.
📊 Track: A separate ESP32-S3 AMOLED display shows live video and each cat’s daily count. The system also saves event clips with audio.
🐾 In a home setup with four cats, @emwstudio reports 89 manually checked drinking events over three days, with all cat identities correctly recognized.
A practical example of complementary models at work: an intelligent model handles frequent detection, while a larger model handles identity confirmation.
We love seeing our community turn MiniCPM’s visual understanding into creative tools for everyday life. What would you build? 💙
🎬 Watch @emwstudio’s demo on Instagram:
instagram.com/reel/De9kUrooQ…
🛠️ Explore @emwstudio’s project:
github.com/emwstudio/esp32-e…
🚀 Build with MiniCPM-V:
GitHub: github.com/OpenBMB/MiniCPM-V
🤗 Hugging Face: huggingface.co/openbmb/MiniC…
#MiniCPM #OpenBMB #ESP32 #OpenSource
