TrepHub

Qwen3.8-27B has been compressed from 55.6GB to just 5.9GB using Bonsai 2. That’s an 89.4% reduction in model size, while retaining a reported 98.2% of aggregate benchmark performance across 20 benchmarks, according to PrismML. The compression uses ternary quantization. Instead of storing weights across a wide range of numerical values, it maps them to just three states: -1, 0 and +1, with scaling factors to preserve useful information. The interesting part is how much capability survives that reduction. PrismML reports that Bonsai 2 also improves performance in areas such as coding agents, multimodal reasoning and long-horizon tool use. Its previous compression of the same 27B model retained around 95% of benchmark performance, making this a further improvement. A 5.9GB model file could make Qwen3.8-27B accessible on more modest hardware, including setups that previously couldn’t accommodate its full weights. There are two important caveats, though. First, 5.9GB is the compressed model size, not the total memory required to run it. Runtime memory still depends on the inference engine, context length, KV cache and other overhead. Second, 98.2% is an aggregate across benchmarks. It doesn’t mean every task retains exactly 98.2% of the original model’s capability. Still, compressing a 27B model this aggressively while preserving most of its measured performance is a notable result. The next question is how it performs in real-world coding, reasoning and agent workflows.

keeping up with AI is a full-time job. that's why every company needs a frontier team: the frontier team's job is to keep up as models change. everyone else's job is to use what they learn we made one @Every and it's completely changed how we work: 0:00 Introduction: Why every company needs a frontier team 2:42 Why the best frontier teams have pirates and architects 4:57 How Head of Evals @hammer_mt builds personal benchmarks 6:46 How Opus edited an explainer video using Slack messages and meeting notes 10:20 Working with AI means protecting other people’s attention 12:26 Why a frontier team needs a “wrangler” like Arielle Shipper 18:58 Why senior editor @jackcheng built a tabless browser 24:37 How resident writer @kplikethebird teaches her AI writing plugin who the reader is 30:51 Inside engineer @JannikJung’s personal AI orchestrator 35:41 Compound engineer @kieranklaassen uses AI to build power tools

defining a new category: phone-use an mcp/api that gives ai agents their own phones cloud phones. real androids. real iphones. tell your agent what to do, and it can operate a phone just like you do. today, ai agents face three fundamental problems and phoneuse solves them 🧵