Videos

Loose Box's $5,999 AMD-based system claims 3.4x faster inference on Deepseek V4 Flash than Nvidia's DGX Spark, but...

A 300B parameter model can run on a 128GB mini PC using aggressive quantization to ~2-3 bits per weight, sparse...

CPU can be faster than GPU for token generation in AI inference, with Xenolith Engine achieving 18.4 tokens per...

Strata, an Engine 1 developer's inference engine for Qwen-3 8B Flash Next, claims 44.8 tokens/second on a 12GB RTX...

Impeccable, Paul Bakaus's Claude skill for design work, is built like a mixture-of-experts system: two sub-agents...

Running DeepSeek V4.1 Flash locally requires at least 510 GB of storage and a multi-GPU setup costing tens of...

DeepSeek V4.1 Flash is a 552B parameter MoE model with only 8B/16B active parameters, matching Opus 5 and GPT 5.6 on...

K2 Horizon MUA's mixture-of-value-experts architecture routes 4 of 64 value experts per word, yielding 4B active...

Tencent's HY4 Preview, a 770B-parameter MoE with 49B active parameters and a 1M-token context, is already generating...

GLM 5.3 Flash is a Chinese MoE model served on domestic chips that costs 10 cents vs $7.50 for comparable...

Qwen3.8-Flash-Next's 51B 'engram' parameters are a lookup table designed to live in slow host memory, not on the...

Qwen3.8 Flash Next, a preview of the Qwen 4 architecture, demonstrates exceptional vision and code generation...

Turbo Field Fair runs a 26B parameter Mixture of Experts model on a Mac with only 2GB of RAM by keeping the shared...

Ornith 1.5 35B-A3B, a Mixture of Experts model based on Qwen 3.5, shows impressive capability for its size,...

GLM-5.3 is the same neural network as its predecessor, but post-training on synthetic environments turned it into...

Nvidia's Nemotron 3.5 Lightning is 88% not a transformer — it's a hybrid of Mamba state-space layers and...

Ling 3.0 Tiny, a 7.9B parameter mixture-of-experts model with 1.3B active parameters, impresses in coding tests...

Waste is a 6,000-line C engine that runs the 2.78-trillion-parameter Kimi K3 mixture-of-experts model on a laptop...

Poolside Laguna S2.1 is a 118B parameter Mixture of Experts model (8B active) that excels at creative writing and...

Kimi K3 nearly matches Anthropic and OpenAI in benchmarks, but its real impact is through architectural innovations...

Thinking Machines' Inkling model is a mid-range open-weight LLM that falls short of cutting-edge performance but...

Tencent released the full version of Hy3, a 295B parameter mixture-of-experts model with 21B active parameters and a...

Tencent HY3, now fully released under Apache 2.0, is a 295B parameter mixture-of-experts model with 21B active...

Hermes agent's latest update introduces a Ministry of Experts feature that uses a panel of models orchestrated by a...

Liquid AI's LFM2.5-8B-A1B is an 8-billion-parameter mixture-of-experts model with 1 billion active parameters,...

NVIDIA's Nemotron-3 models come in Nano, Super, and Ultra variants sized for consumer hardware, workstation/server...