Benchmark results for Bonsai Q2_0 and Q1_0 on my GX10, using strict EvalPlus: Ternary Bonsai Q2_0 • HumanEval+: 152/164 92.68% • MBPP+: 305/378 80.69% • Total Plus solved: 457/542 Bonsai Q1_0 • HumanEval+: 149/164 90.85% • MBPP+: 282/378 74.60% • Total Plus solved: 431/542 Both completed all 542 generations with zero errors. Ternary recovered 26 tasks over Q1_0. Of those, 23 came from MBPP+ and only 3 from HumanEval+. For context, PrismML reports Qwen3.6-27B FP16 at 95.12% HumanEval+, 83.33% MBPP+, and 88.74% across its coding category including LiveCodeBench. That's a very very solid model for such weights and considering its file size.

i came across this llama.cpp fork that uses less VRAM and KV memory. on my GX10, Qwen 3.5-family Bonsai 27B, 60k context, median of three runs: q8_0: • 2,006 MiB KV • 792 tok/s prefill • 27.2 tok/s decode KVarN5 + 1,024-token tail: • 1,349 MiB KV • 740 tok/s prefill • 25.6 tok/s decode The trade: 32.8% less KV memory for roughly 6% lower throughput might be a solid option for people trying to squeeze in some more tok/s ! repo:

Just weeks after the release of Luna Abyss, the entire development team at Kwalee Labs has reportedly been laid off. Luna Abyss launched on May 21 after about seven years of development. The team, which originally began as Bonsai Collective before joining Kwalee, had around 25 developers at its peak. The game received positive reviews and has a Metacritic score of 80.

♥️Huge thanks to @nvidia for the DGX Sparks for @openclaw engineering team. We have been putting these to hard work, product announcements soon with our joint efforts to improve local model experience and empowering use-cases locally beyond what is currently possible on OpenClaw. I've been particularly enjoying newer local models like Bonsai and Poolside Laguna and distributed inference with tools like Exo.