Michael Jentsch @msoftware

🚀 Introducing Gemini-SQL2, our breakthrough text-to-SQL capability powered by Gemini 3.1 Pro! We've achieved state-of-the-art results on the highly competitive BIRD benchmark, translating natural language into execution-ready SQL queries. 🧵👇

Cohere just released North Mini Code, a small 30B parameter (3B active) open weights coding model that scores 27.6 on the Artificial Analysis Intelligence Index Less than a month since @cohere's last model release, Command A+, has launched another open weights model that is optimized for coding, and much smaller at 30B total parameters and 3B active parameters. Key Takeaways: ➤ Achieves 27.6 on the Artificial Analysis Intelligence Index, above gpt-oss-20B (high) at 24.5 and just below Mistral Small 4 (119B parameters, 6.5B active) at 27.8 ➤ Scores competitively on the Artificial Analysis Coding Index (weighted average of Terminal-Bench Hard and SciCode) against open weights models in its size class, scoring 33.4, significantly above GLM-4.7-Flash at 25.9, and below Qwen3.6 35B A3B at 35.2. However, it underperforms on non-coding agentic tasks, scoring 14% on GDPval-AA and 37% on 𝜏²-Bench Telecom ➤ On Cohere’s API, North Mini Code is faster than several comparable open weights models of its intelligence and size class (~199 output tokens per second) ➤ North Mini Code is a text-only 30B total parameter and 3B active parameter model, and is open-sourced under the Apache 2.0 license

Building super fast experiences with Gemma just got easier. Gemma 4 MTP is now officially merged into llama.cpp. Developers can now pair MTP with Gemma 4 QAT for a fast, lightweight setup.

Building a custom chatbot can be tricky, especially if you want to keep the info private & efficient. So here, @atuo_elabonga teaches you how to create a local Retrieval-Augmented Generation application using Ollama & ChromaDB in R. By the end, you'll have a Shiny interface for your chatbot that fetches relevant info while staying private. freecodecamp.org/news/build-a-l…

More Gemma 4! New QAT Gemma 4 checkpoints with similar performance while using ~4x less memory! It comes with a new mobile quantization format that reduces memory footprint of Gemma 4 E2B to just 1GB. Quantization-Aware Training (QAT) simulates low-precision operations during training to allow loss-less quantization afterwards for smaller, faster models while maintaining accuracy. Available on @huggingface and directly runnable.

Google’s newly released open weights model, Gemma 4 12B, supports transcription but is far from the frontier, scoring 8.8% on AA-WER (#58) Gemma 4 12B is the latest release from @GoogleDeepMind in the Gemma 4 family. With a score of 8.8% on AA-WER, it is able to capture a reasonable amount of conversation context, but underperforms compared to transcription-focused open weights models like Voxtral Mini Transcribe 2 (3.6% WER, with 4B parameters) and slightly larger open weights language models like Voxtral Small (2.8% WER, with 12B parameters). The new model launched alongside their local dictation app, Eloquent, available on MacOS and iOS. Gemma 4 12B is the largest in the Gemma 4 family to support transcription, alongside Gemma 4 E4B and Gemma 4 E2B, with Gemma 4 31B and Gemma 4 26B A4B supporting text, image and video input only. These models are available on a variety of platforms including Hugging Face, Ollama and LMStudio. We are currently running Gemma 4 12B through the full Artificial Analysis Intelligence Index and will share results soon.

LiteRT-LM support for Flutter is coming soon to the flutter_gemma package 💖 This will enable you to run powerful on-device AI models like Gemma 4 across devices, all while helping to ensure peak performance with GPU and NPU acceleration across all of Flutter’s stable platforms.

MiniMax-M3 is now the best open-source model on the AAI It's ahead of Kimi-K2.6, GLM-5.1 and DeepSeek-V4-Pro On the Coding Index it's slightly behind V4-Pro and K2.6, but it's beating them on the Agentic Index CritPt scores are a bit worrisome. But the biggest highlight here is probably its reasoning efficiency compared to other open models. (but still miles behind closed models)

keyboard_arrow_left Previous keyboard_arrow_right Next

むむ？この4Bモデルのスコアの跳ね方が異常。試す huggingface.co/Hcompany/Holo-… まさかQwen3.5−397b−a17bに匹敵するなぞ

Higgs Audio v3 TTS is here. Built for voice AI that speaks, not just reads: • 100 languages with single-digit WER/CER • inline control over emotion, style, prosody, and sound effects • API, Workspace, and open weights • Blog 👉 boson.ai/blog/higgs-aud… Watch the demo 👇

Key features: • Free historical data • Includes walk forward analysis • A super-fast backtesting engine • Create and execute trading rules across different assets Get it here: pybroker.com/en/latest/

Meet Gemma 4 12B! A unified, encoder-free multimodal model designed to bring high-performance intelligence directly to your laptop, and released under an Apache 2.0 license. Bridging the gap between edge efficiency and advanced reasoning. Here is what’s new with Gemma 4 12B: 👇

Dinosaurs: Surrealism, Renaissance, Cubism, and more.... By Noemi Fadda, Used with permission, displate.com/displate/596473

The Twelve Years of Creality: AI Ecosystem 12 years of creating with Makers worldwide — thank you for being part of the journey. Today is Filament Explorer Day For Makers who love testing materials, fine-tuning settings, and discovering the full potential of every filament, creativity starts before the print. With the Creality ecosystem, every spool becomes a tool for turning ideas into reality. To celebrate our passionate Filament Makers, we’ve prepared a special giveaway. Giveaway Prize: Choose Any 4 Rolls of Creality Filament How to Enter: Follow @crealityfilament & @Creality3dP Like this post Share this post to your story for a higher chance to win Winners will be selected across all social media platforms on June 8th. Discover More: to.store.creality.com/12th-anniversa… Thank you for creating with Creality — and cheers to the next chapter of making together. #12YearsOfCreality #MakeWithCreality #crealityfilament #filament #3DPrinting

The World's full open source Smallest Desktop-Scale Two-Wheeled Robot. Github: MuShibo/Micro-Wheeled_leg-Robot

🌞This is big Local AI news! A new open-source Computer-Use LLM has just launched. Holo 3.1 is H Company’s (🇫🇷) new local computer-use agent model that beats Qwen3.5-397B, Kimi-K2.5, and Sonnet 4.6! Since it is built for local deployment → ⬩ Runs fully on your machine (MacBook, Windows PC, DGX Spark, RTX Spark) ⬩ Based on Qwen architecture, specialized for GUI understanding & computer control ⬩ Optimized checkpoints: NVFP4, FP8 & Q4 GGUF (0.8B to 35B sizes) ⬩ Strong gains: 79.3% on AndroidWorld benchmark (35B model) 💻 Comparison to Qwen3.5: Holo 3.1 is fine-tuned specifically for computer-use agents (screen understanding, planning, clicking, navigation). Better at real GUI tasks than general-purpose Qwen3.5, especially when running locally.⚡

H @hcompany_ai

2 weeks ago

Computer-use agents are moving from the cloud to your local machine. Fast. When we launched Holo3 two months ago, the production feedback was clear: digital agents need to be blazing fast, cost-effective, and versatile. Today, we're dropping Holo 3.1, engineered to run

keyboard_arrow_left Previous keyboard_arrow_right Next

35 72 505 231K 365

Qwen3.6 35B A3B can't fill out a paper form on its own. But give it NVIDIA's LocateAnything-3B — the #1 trending model on HuggingFace — as its eyes, and the two small models get it done together. (The test: place each element at the right pixel position on a blank form image, not type into a field.) Setup: > Qwen is the brain (main model), LocateAnything is the eyes (helper model acting as a tool). > I gave Qwen a new tool: ask "where's the email field?" and LocateAnything returns the exact x, y, width, height. > The blue boxes on the screen are its detections. Look how tight they are — it nails every field. Result: > Qwen3.6 35B A3B + LocateAnything-3B: form completed, all info correct. > Name, DOB, ID, gender, marital status, nationality, email, phone, address, postal code: all landed in the right field areas. > Character-box alignment still a touch loose, but every value is where it belongs. > 9m10s, 224.5k input, 24.3k output, 21 turns. Why it matters: > Qwen alone can't finish this test. Bolt on a 3B model that does exactly one thing > locate > and suddenly it can. > A combination of small models can do the work of a single large one.

MOSS-TTS-v1.5 just reached #1 on Hugging Face Trending for Text-to-Speech, with 20.6K downloads. A multilingual, controllable TTS model with stable voice cloning, long-form generation, and precise pause control. MOSS-TTS-v1.5 is now officially supported by vLLM-Omni and SGLang-Omni. Built by OpenMOSS-Team. Try it: GitHub: github.com/OpenMOSS/MOSS-… Hugging Face: huggingface.co/OpenMOSS-Team/… ModelScope: modelscope.ai/organization/o…

Seven new models launching at Build: let’s go! Reasoning. Code. Image. Transcribe. Voice. Built from scratch on a clean data lineage, designed for efficiency, working seamlessly as a family of models Thread 🧵 #MSBuild

Most neural nets are still based on the model of a neuron as proposed in the 1950's: u = activation(w·x + b) In a new paper, researchers propose a more accurate model of a biological brain neuron and found that it has quite a few advantages, like needing less training data.