Qwen3.6 27B Fits Into 12GB VRAM
immersive-matthew · reddit · 2026-07-15
A Reddit post introduces a new Ternary version that compresses Qwen3.6 27B to a smaller size, allowing it to run locally on 12GB VRAM GPUs.
- The author states this Bonsai 27B version is updated from Qwen3.6 27B, featuring a smaller file size and significantly reduced memory usage, officially requiring only about 1/10 of the original FP16 model's memory.
- The tradeoff is slightly increased computational overhead, but it reportedly retains around 95% of the original model's benchmark performance.
- The post also mentions that the Binary version is small enough to fit into mobile devices and includes a YouTube tutorial to help users easily deploy it locally.
Related event: Bonsai 27B: The first 27B model that runs on phones(15 posts)→
More from Infra
- Anthropic hits $16.3B H1 revenue as compute efficiency jumps 5.2x in 18 months — daniel_mac8 · 2026-10-27
- Tenstorrent runs Proteina protein model at high throughput, claims no HBM shortage — gajesh · 2026-10-27
- YC-backed Hopper launches voice-agent inference cloud: 80 ms TTFT at $0.30/1M tokens — ycombinator · 2026-10-27
- A100 rental prices still rising in 2026 per a16z data, Jevons paradox hits compute — VraserX · 2026-10-27
- Anthropic S-1 reveals $800B+ in off-books future compute commitments — GaryMarcus · 2026-10-27
- ~42% of Anthropic's ~11GW contracted compute is NVIDIA GPUs — ns123abc · 2026-10-27