MiniCPM-V 4.5 beats much larger models by focusing on efficiency instead of brute force. Instead, it combines (1) extreme video compression, (2) smarter OCR/document pretraining, (3) controllable fast/deep reasoning, and (4) efficient alignment. So an 8B model can outperform
OPEN SOURCE
-
Limited Control and Understanding in Model Fine-tuning Approaches
By
–
Maybe it's a n-of-1 problem, but there's very little control (hyperparams etc.), and very little understanding (both on HOW it's being done (is it LoRA, fft, etc.) and telemetry-wise). May be n-of-1 b/c most people who want this stuff will run their own shit, but I personally
-
100+ Free Open Source AI Agents and RAG Tutorials
By
–
Stay tuned for more such interesting posts → @Saboo_Shubham_ I have created 100+ AI Agents and RAG tutorials, 100% free and opensource. P.S: Don't forget to star the repo to show your support
-
MiniCPM-V 4.5 Open Source Multimodal LLM Outperforms GPT-4o
By
–
GPT-4o level multimodal LLM running on your phone.
— Shubham Saboo (@Saboo_Shubham_) 5 septembre 2025
MiniCPM-V 4.5 outperforms GPT-4o, Gemini-2.0 Pro, and Qwen2.5-VL 72B on vision and language AI tasks.
It can even understand videos and perform OCR in 30+ languages.
And it's 100% Opensource. pic.twitter.com/wY70HlmzDSGPT-4o level multimodal LLM running on your phone. MiniCPM-V 4.5 outperforms GPT-4o, Gemini-2.0 Pro, and Qwen2.5-VL 72B on vision and language AI tasks. It can even understand videos and perform OCR in 30+ languages. And it's 100% Opensource.
-
How to Build AI Agents Without Buying API Keys
By
–
Wait… so I can build AI agents without buying API keys??
-
Token Origin Tracking and Fair Trade Implementation Updates
By
–
Now tracking the origin of each parsed token to ensure it's organic and Fair Trade, and the location of tokens is shown in better error messages… Many other changes today, will likely make the next release tomorrow!
-
EmbeddingGemma: Multilingual On-Device Embedding Model in Browser
By
–
model carrot vibe coding another Google EmbeddingGemma app, a sota multilingual embedding model perfect for on-device use cases using transformers.js and onnx-community/embeddinggemma-300m-ONNX At only 308M params, the model can run 100% locally in your browser app:
-
Adam Optimizer: Fast.ai’s 2018 Breakthrough Research Discovery
By
–
Guess who was the 1st to point out that Adam can be used for pretty much everything? (Answer: it was @fastdotai back in 2018 — @GuggerSylvain
's first research project in fact!) https://
fast.ai/posts/2018-07-
02-adam-weight-decay.html
… -
Liquid AI Japan 350M Model Rivals GPT-4o Translation Performance
By
–
Liquid AI Japan cooked with this 350M param model on par with GPT-4o for English Japanese translation That's a really nice example of fine-tuning done right