# Bonsai 27B Hacker Approved page: https://hackerapproved.com/p/bonsai-27b/ 45 people recommended Bonsai 27B in Hacker News comments, 22 had mixed feelings and 41 criticized it, November 2025 to October 2026. It was recommended in 8 of 12 months, in 21 threads. It does not meet the Hacker Approved badge rule. **Verdict:** People recommend it as a 27B-class model that fits in little memory and runs on laptops and phones. The common caveat is that it weakens on long tasks and on vision. Summary written by Claude from 51 comments. It was written before the November 2026 recount, and 42 of those comments still count. ## Why people recommend it - Fits in little memory (5 comments in the sample): "My brief experiments with the ternary version suggest that it broadly meets their claim to be a 27B model that fits in much less RAM…" ([source](https://news.ycombinator.com/item?id=48914973)) "this new Bonsai is nice because the original was already exciting for making use of low VRAM devices" ([source](https://news.ycombinator.com/item?id=49748455)) "You can run bonsai2 27b on basically any gaming GPU from the last 15 years" ([source](https://news.ycombinator.com/item?id=50266151)) - Strong at math and coding (3 comments in the sample): "this model crushes Gemma 4 12b on math and coding, is slightly worse on knowledge and tool calling, and is significantly worse on vision tasks" ([source](https://news.ycombinator.com/item?id=48913981)) "it is quite an impressive model! I've been grilling it on some deep learning/computer vision stuff and it's aced everything so far." ([source](https://news.ycombinator.com/item?id=48915504)) "shockingly smart for a tiny model (but also not Good Enough, by my above definition" ([source](https://news.ycombinator.com/item?id=47783954)) - Strong size-to-performance tradeoff (1 comment in the sample): "quantization and pruning make the model better at the size-performance pareto" ([source](https://news.ycombinator.com/item?id=48915992)) - Good for classification and summarization (1 comment in the sample): "They’re fairly useful as constrain domain classifiers due to the low memory requirements" ([source](https://news.ycombinator.com/item?id=49757104)) ## What people warn about - Breaks down on long tasks (2 comments in the sample): "The most common failure mode is that it goes into loops and you come back to find it exhausted the output length" ([source](https://news.ycombinator.com/item?id=49755542)) "it's amazing they work as well as they do. Use it for any longer task and they fall apart spectacularly" ([source](https://news.ycombinator.com/item?id=49747164)) - Needs special runtime or WebGPU (2 comments in the sample): "The experience is vastly improved but still requires a custom llama.cpp fork to use as of right now." ([source](https://news.ycombinator.com/item?id=49956802)) "The online demos require WebGPU so Firefox on mobilr and privacy enhanced browsers will break." ([source](https://news.ycombinator.com/item?id=48347673)) - Claims overstated or cherry-picked (1 comment in the sample): "The benchmarks they chose are rather cherry picked to not include long context or difficult ones" ([source](https://news.ycombinator.com/item?id=49749762)) - Weaker on world knowledge and vision (1 comment in the sample): "The V1 was noticeably weak on world knowledge but here the 3.8 base model is geared more towards reasoning than world knowledge anyway" ([source](https://news.ycombinator.com/item?id=49748117)) ## Other open-weight language models people recommend - [Qwen 3.8 27B](https://hackerapproved.com/p/qwen-3-8-27b/): 804 recommend, 242 mixed, 301 criticize - [Z.ai GLM](https://hackerapproved.com/p/glm-5-2/): 750 recommend, 181 mixed, 278 criticize - [Kimi](https://hackerapproved.com/p/kimi-k3/): 652 recommend, 182 mixed, 275 criticize - [Gemma 4](https://hackerapproved.com/p/gemma-4/): 524 recommend, 120 mixed, 166 criticize ## The badge A product earns the Hacker Approved badge when at least 10 people recommend it, recommenders outnumber critics at least 2 to 1 (mixed feelings count toward neither side), it was recommended in at least 4 of the 12 months and in at least 5 different threads, and no single thread supplies more than half of the people who recommend it. Health and money products are never badged. ## Cite as Hacker Approved, "Bonsai 27B", https://hackerapproved.com/p/bonsai-27b/ (45 recommend, 22 mixed, 41 criticize; Hacker News comments, November 2025 to October 2026). Counts updated 2 Nov 2026, with data through October 31, 2026.