AI Pulse by Inblix

Topic: mixture-of-experts

14 articles

Explore our coverage of mixture-of-experts — 14 curated articles, summaries, and related resources from the Inblix archive.

Alibaba's 2.4T param Qwen3.8-Max dethrones DeepSeek on Arena.AI, but at 14x the cost — Inblix summary
Industry

Alibaba's 2.4T param Qwen3.8-Max dethrones DeepSeek on Arena.AI, but at 14x the cost

AI News · Aug 5, 2026 · 2 min read

The race to build the biggest—and cheapest—AI model in China just got a new frontrunner, depending on what you're measu...

Cursor drops MoK megakernel: 2.37x faster MoE training, but it demands a $2M rack — Inblix summary
AI News

Cursor drops MoK megakernel: 2.37x faster MoE training, but it demands a $2M rack

MarkTechPost · Aug 4, 2026 · 3 min read

Cursor just open-sourced the secret sauce behind its Composer models, and it’s not for the hobbyist crowd. The Mixture-...

Inkling-Small beats its 975B teacher on reasoning, runs on a single GPU — Inblix summary
AI News

Inkling-Small beats its 975B teacher on reasoning, runs on a single GPU

MarkTechPost · Aug 2, 2026 · 2 min read

Thinking Machines Lab just flipped a familiar script. Usually, the smaller, distilled model trails the giant teacher. I...

AMD ships a 16B-param MoE model trained entirely on its own GPUs, but the license is research-only — Inblix summary
AI News

AMD ships a 16B-param MoE model trained entirely on its own GPUs, but the license is research-only

MarkTechPost · Aug 1, 2026 · 2 min read

AMD just dropped a Mixture-of-Experts model designed to prove a point: you can train a competitive large language model...

Microsoft's new AI defense model hits 95.95% bug-finding rate but can't exploit a thing — Inblix summary
AI News

Microsoft's new AI defense model hits 95.95% bug-finding rate but can't exploit a thing

MarkTechPost · Jul 28, 2026 · 2 min read

Microsoft just dropped its first model purpose-built for cyber defense, and the design choices say a lot about where AI...

Induction Labs’ 106B model masters tasks from raw video, no action labels needed — Inblix summary
AI News

Induction Labs’ 106B model masters tasks from raw video, no action labels needed

MarkTechPost · Jul 26, 2026 · 2 min read

The standard recipe for training AI agents from video has a stubborn requirement: you need to know the action that caus...

Laguna S 2.1 found a proof for a 1975 math problem — and it's an 8B model — Inblix summary
AI News

Laguna S 2.1 found a proof for a 1975 math problem — and it's an 8B model

The Decoder · Jul 23, 2026 · 2 min read

Poolside just dropped Laguna S 2.1, and the most telling detail isn't a benchmark score — it's that the model independe...

Poolside's Laguna S 2.1 crushes DeepSeek on DeepSWE with one-sixth the active parameters — Inblix summary
AI News

Poolside's Laguna S 2.1 crushes DeepSeek on DeepSWE with one-sixth the active parameters

MarkTechPost · Jul 22, 2026 · 2 min read

Poolside dropped Laguna S 2.1 on Thursday, a 118-billion-parameter open-weight coding model that punches wildly above i...

China's Moonshot Just Dropped a 2.8T Parameter Model That Trades Compute for Memory — Inblix summary
Industry

China's Moonshot Just Dropped a 2.8T Parameter Model That Trades Compute for Memory

AI News · Jul 20, 2026 · 2 min read

The Kimi K3 from Moonshot AI isn't just another large language model; at 2.8 trillion parameters, it's the largest open...

Thinking Machines drops Inkling: a 975B-param open model that sees, hears, and reads — Inblix summary
Research

Thinking Machines drops Inkling: a 975B-param open model that sees, hears, and reads

Hugging Face Blog · Jul 15, 2026 · 2 min read

Forget the single-mode giants. Thinking Machines just put Inkling on Hugging Face, and it’s not just another large lang...

Soofi S: The 30B German moe model that spanks 70B rivals — Inblix summary
AI News

Soofi S: The 30B German moe model that spanks 70B rivals

The Decoder · Jul 13, 2026 · 3 min read

A German research consortium just dropped Soofi S, and the benchmarks make a loud statement: you don't need a massive,...

OpenAI Ships Open-Weight Reasoning Models, First Since GPT-2 — Inblix summary
Product

OpenAI Ships Open-Weight Reasoning Models, First Since GPT-2

OpenAI Blog · Jul 13, 2026 · 2 min read

OpenAI just did something it hasn't done since 2019: release an open-weight language model. The gpt-oss family lands wi...

Transformers 4.49 quietly fixes the mess that made MoE models a deployment nightmare — Inblix summary
Research

Transformers 4.49 quietly fixes the mess that made MoE models a deployment nightmare

Hugging Face Blog · Feb 26, 2026 · 2 min read

The dirty secret of Mixture of Experts models has been deployment. For the past year, while everyone celebrated models...

Hugging Face just made GPT-OSS's best tricks part of every model — Inblix summary
Research

Hugging Face just made GPT-OSS's best tricks part of every model

Hugging Face Blog · Sep 11, 2025 · 2 min read

Remember how GPT-OSS got those impressive efficiency numbers? The secret wasn't just the architecture — it was a stack...