atomic.chat
@atomic_chat_hq
The simplest way to run open models on any hardware. Powered by TurboQuant. Team: @gladkos @skinbagwbones @AlexFromAtomic @danyurkin @quantizedden
68 Following 18.9K Followers
Run Nemotron 3.5 Super VL locally on 64GB Mac 💻
We released 7 quants of @NVIDIAAI's new 120B vision model on Hugging Face from 131.6GB down to 59.3GB
For a 64GB Mac we recommend the 59.3GB AD-IQ2_M with a top-1% of 89.8% compared to BF16
GGUFs 👇
Show more
Run UNCENSORED Qwen-Image-2.1-Turbo locally 🔥
Our Atomic Dynamic GGUF quants ladder starts from 3.3 GB to 16.4 GB
Refusal dropped to 88.9% to 1.2%
Get quants here👇
Local Laya blitzed other models on RTX 4090 ⚡️
Four local decision models raced down Wikipedia picking out every word that names a centipede and Laya took 3.9 ms a word so it ended up being the fastest overall
Run decision models locally ->
Show more
介绍比Jev快50倍，在你设备上跑的laya-mlx!
只在你的设备上占用最高1G内存
Laya是一个开源的类似于Jev的,基于文本输出概率的分类系统
我将其移植到MLX,并且做了一些性能优化!
视频中就是这个模型在我的本地M3Max上玩贪吃蛇
这个模型能够以每秒决策60次的速度玩贪吃蛇！
Show more
Run Liquid AI's d1 decision models locally ⚖️
We dropped GGUF quants of d1-3B and d1-omni-600M on Hugging Face, from 256 MB to 5.4 GB
The sweet spot is our 1.66 GB d1-3B AD-Q4_K_M and it's top-1% is 97.1% compared to BF16
GGUFs:
Show more
Today we release Open d1: two open-weight multimodal models in our d1 decision model family.
> d1-3B: text + vision
> d1-omni-600M: text + image or text + audio
> Real-time decision making anywhere, from data centers such as @nvidia DGX to RTX workstations to Jetson at the edge.
1/
Show more
Claude Haiku 5.5 cost 12x more than GPT-6 Luna for the same scene 🐉
Both have the same API price, yet building the same animated voxel 3D pagoda island cost $24.96 on Claude Haiku 5.5 and $1.91 on GPT-6 Luna
Run both models via API ->
Show more
Local Liquid d1 3B crushed OpenAI Decision at Snake 🐍
@liquidai d1 on a single RTX 3090 decides in 8 ms so its snake made 82 moves a second and ate 86 apples while OpenAI's API took 345 ms a call and ate 3
Run decision models ->
Show more
Today we release Open d1: two open-weight multimodal models in our d1 decision model family.
> d1-3B: text + vision
> d1-omni-600M: text + image or text + audio
> Real-time decision making anywhere, from data centers such as @nvidia DGX to RTX workstations to Jetson at the edge.
1/
Show more
Local Laya beat @OpenAI's decision API 🕷
Four decision models raced down a web page ripping out links that name a spider and Laya running locally on an M5 Max took 34 ms a link so it checked 3x more than the cloud models
Run decision models locally ->
Show more
Let your app choose the right model, tool, or action in near real-time with Decisions API, now available to all developers in public beta.
The Decisions API makes decisions up to 10x faster than GPT-6 Luna through the Responses API.
Show more
Le Chonk enters the Frontier race 🐈
We put @MistralAI Large 4, MiMo V2.6 Pro, Opus 5.5 and GPT-6.1 Sol side by side with voxel, ASCII, Blender and SVG remakes of the launch clip
Run AI models locally and via API ->
Show more
Meet Mistral Large 4, aka Le Chonk.
• 1T parameters, natively multimodal. 49B active.
It is the best open weights model from US or Europe on aggregated benchmarks.
• State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding.
• Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure.
• Available to all via API today. Working with cybersecurity partners privately.
Open weights release end of October.
Show more
TensorRT 206% faster than llama.cpp ✳️
We wired @NVIDIAAI's TensorRT into Atomic Chat's engine and tested Qwen3.5-4B/Qwen 3-8B via CUDA on a Blackwell RTX 5090, measuring 3.43s vs 10.49s with llama.cpp
Run AI models locally ->
Show more
What could your app do with local AI? 🤔
Turn speech into text, select objects in video, or create images from a prompt. Our latest blog walks through open-source C++ samples that bring these capabilities to @Windows and Linux with NVIDIA TensorRT RTX.
Show more
Laya plays Plants vs Zombies on CPU 🧟
We patched the llama.cpp engine to run Laya natively so it works on a regular CPU with no GPU and here a fine-tuned Laya makes every move until it beats the Plants vs Zombies pool level
Run decision models locally->
Show more
介绍比Jev快50倍，在你设备上跑的laya-mlx!
只在你的设备上占用最高1G内存
Laya是一个开源的类似于Jev的,基于文本输出概率的分类系统
我将其移植到MLX,并且做了一些性能优化!
视频中就是这个模型在我的本地M3Max上玩贪吃蛇
这个模型能够以每秒决策60次的速度玩贪吃蛇！
Show more
Run decision models with our llama.cpp engine ⚙️
We're bringing support for over 10 System One models into Atomic Chat, now your device makes choices 10.43x faster than Jev's API utilising only the CPU
Run decision models locally ->
Show more
介绍比Jev快50倍，在你设备上跑的laya-mlx!
只在你的设备上占用最高1G内存
Laya是一个开源的类似于Jev的,基于文本输出概率的分类系统
我将其移植到MLX,并且做了一些性能优化!
视频中就是这个模型在我的本地M3Max上玩贪吃蛇
这个模型能够以每秒决策60次的速度玩贪吃蛇！
Show more
Run Uncensored Qwen-Image 2.1 on 16GB Mac
@Alibaba_Qwen's image model with no filters makes the images cloud apps refuse and your prompts never leave your Mac, all of it done locally
Run AI models locally ->
Show more
Meet Qwen-Image-2.1, the most balanced and cost-effective image generation model in the Qwen-Image series! Now open weights! 🎨
A unified model for both generation and editing, delivering top-tier quality in a lightweight package.
Highlights: 👀
- Compact & exceptionally fast: A lightweight 7B architecture that outperforms most closed-source models, with drastically accelerated inference for multi-image inputs.
- Native transparency: Natively generates and edits RGBA layers, enabling seamless compositing and text editing within transparent images.
- Versatile, high-fidelity editing: Supports up to 10 reference images and precise local control while preserving strict fidelity for portraits and products.
- Broad coverage & stunning aesthetics: Excels at panoramas, infographics, and virtual try-ons, delivering realistic textures and elegant typography.
Start to create your next masterpiece with Qwen-Image-2.1! 🖼️
- Blog:
- GitHub:
- Model Scope:
- Hugging Face:
Show more
Atomic Fusion cut GPT-6.1 Sol’s API bill by 77% 💥
We tested cloud-only, local-only and orchestrated builds. Cloud GPT-6.1 made a plan and orchestrated local Qwen 3.8 27B, making it cheaper with minimal quality loss
Try Atomic Fusion ->
Show more
GPT 6.1 Sol 30x cheaper than Sonnet 5.5 at 3D ⚙️
openAI just dropped GPT 6.1 Sol, we put it up against Sonnet 5.5 and made them create a 3D keyboard assembly video in @threejs
Run both models via sub ->
Show more
GPT-6.1 Sol: near-Astra intelligence for a fifth of the price.
It’s the most cost-efficient model for its performance available today.
Opus 5.5 vs Sonnet 5.5 vs GPT 6 Sol 3D battle
@claudeai just dropped Sonnet 5.5 and we put it against recent models by making each one create a steampunk 3D whale in Blender, after that animate and texture it
Run cloud models via API and sub ->
Show more
Introducing Claude Sonnet 5.5, the second model in the Claude 5.5 family.
It’s a clear upgrade over Sonnet 5, runs more than 30% faster, and costs up to 30% less for most work.
Show more
Run LFM2.5-VL-3B on 16GB Mac with DSpark ⚡️
@liquidai's new DSpark drafter delivers up to 3.13x faster decoding on M5 chips with MLX-VLM in its vision benchmarks while keeping identical outputs under greedy decoding
Run AI models locally ->
Show more
Today, we release an experimental DSpark draft model for LFM2.5-VL-3B, bringing speculative decoding to our vision-language models.
A lightweight drafter proposes multiple tokens ahead, and the target model verifies them together in a single pass. This accelerates generation without changing output quality.
Across six vision-language task categories at batch size 1 and temperature 0:
> MLX on M5 Max: up to 3.13x faster decoding and 2.62x end to end
> llama.cpp on M3 Ultra: up to 2.14x faster decoding and 1.77x end to end
> SGLang on H100: up to 2.66x faster decoding and 2.27x end to end
All evaluations were collected using Pipette, the benchmarking infrastructure behind Liquid AI's public device-performance data.
🧵
Show more
Space Bunny Alpha mogged by DeepSeek V4.1 Flash 🐳
We asked both to build Minecraft in one HTML file and DeepSeek V4.1 Flash made a forest biome you can walk and dig in while Space Bunny Alpha's ground has holes showing the sky
Run AI models locally ->
Show more
Space Bunny (stealth model) is free for the next week
- 1M Context
- Multi-modal
- Zero Data Retention
Run uncensored Bonsai 2 27B on 16 GB MacBook 💻
Our 9.7 MB adapter cuts refusals from 99% to 0% at runtime on both the 5.95 GB and 7.21 GB versions and still answers 73.0% of MMLU questions right vs 73.2% for the original
Get it on our HF ⬇️
Show more
Today, we’re announcing Ternary Bonsai 2 27B.
Based on Qwen3.8 27B, Bonsai 2 27B is 9x smaller than its full-precision counterpart while retaining 98.2% of its aggregate benchmark performance.
Two months after the first Bonsai 27B release, the biggest change is quality. The footprint remains 5.9 GB, but the gap to full precision has narrowed materially, with particularly strong gains in agentic coding, multimodal reasoning, and long-horizon tool use.
Ternary Bonsai 2 27B is available today under Apache 2.0.
Show more
Opus 5.5 destroyed GPT 6 Sol at Blender 3D Pelican 🦩
Both frontier models dropped today so we gave them the same brief to animate a looping 3D pelican on a bicycle in Blender and Opus 5.5 came back with the more charming ride
Run both models via API ->
Show more
Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family.
It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.
Local Laya moggs Jev at @grok 4.7-built Tetris 🧩
An open-weights System One model called Laya, beat cloud-based Jev at playing Tetris by making decisions 11 times faster, running locally on a 16GB MacBook Air!
Run AI models locally ->
Show more
介绍比Jev快50倍，在你设备上跑的laya-mlx!
只在你的设备上占用最高1G内存
Laya是一个开源的类似于Jev的,基于文本输出概率的分类系统
我将其移植到MLX,并且做了一些性能优化!
视频中就是这个模型在我的本地M3Max上玩贪吃蛇
这个模型能够以每秒决策60次的速度玩贪吃蛇！
Show more
