← CanIRunLocal · All hardware · All models

Can I run Bonsai 2 2B (ternary) locally?

Prism ML · 2B parameters · chat, vision, reasoning.

Ollama weight files: Q2_0 0.91 GB (bonsai-2:2b).

Yes on 56 machines in this list. The smallest is M1 8GB: Q2_0, about 2 GB, roughly 60 tokens per second, via Ollama. ollama run bonsai-2:2b

DGX Spark 128GB: Q2_0, ~2 GB, fits in fast memory, ~225 tok/s — ollama run bonsai-2:2b
Mac Studio M5 Max 128GB: Q2_0, ~2 GB, fits in fast memory, ~540 tok/s — ollama run bonsai-2:2b
Mac Studio M5 Ultra 256GB: Q2_0, ~2 GB, fits in fast memory, ~1055 tok/s — ollama run bonsai-2:2b

Setup on DGX Spark 128GB (NVIDIA, ASUS, Dell, Lenovo, HP, MSI, Acer, Gigabyte)

Ollama: ollama run bonsai-2:2b

Open Bonsai 2 2B (ternary) in the checker

Hardware that fits

Hardware	Runtime	Quant	Memory	Fit	Speed	Command
M1 8GB	Ollama	Q2_0	2 GB	fits in fast memory	~60 tok/s	`ollama run bonsai-2:2b`
M2 8GB	Ollama	Q2_0	2 GB	fits in fast memory	~88 tok/s	`ollama run bonsai-2:2b`
M3 8GB	Ollama	Q2_0	2 GB	fits in fast memory	~88 tok/s	`ollama run bonsai-2:2b`
RTX 3060 Ti 8GB	Ollama	Q2_0	2 GB	fits in fast memory	~295 tok/s	`ollama run bonsai-2:2b`
RTX 3070 8GB	Ollama	Q2_0	2 GB	fits in fast memory	~295 tok/s	`ollama run bonsai-2:2b`
RTX 4060 8GB	Ollama	Q2_0	2 GB	fits in fast memory	~179 tok/s	`ollama run bonsai-2:2b`
RX 7600 8GB	Ollama	Q2_0	2 GB	fits in fast memory	~190 tok/s	`ollama run bonsai-2:2b`
RTX 3080 10GB	Ollama	Q2_0	2 GB	fits in fast memory	~501 tok/s	`ollama run bonsai-2:2b`
M1 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~60 tok/s	`ollama run bonsai-2:2b`
M2 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~88 tok/s	`ollama run bonsai-2:2b`
M3 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~88 tok/s	`ollama run bonsai-2:2b`
M4 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~105 tok/s	`ollama run bonsai-2:2b`
RTX 3060 12GB	Ollama	Q2_0	2 GB	fits in fast memory	~237 tok/s	`ollama run bonsai-2:2b`
RTX 4070 12GB	Ollama	Q2_0	2 GB	fits in fast memory	~332 tok/s	`ollama run bonsai-2:2b`
RTX 5070 12GB	Ollama	Q2_0	2 GB	fits in fast memory	~443 tok/s	`ollama run bonsai-2:2b`
RTX 4060 Ti 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~190 tok/s	`ollama run bonsai-2:2b`
RTX 4070 Ti Super 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~443 tok/s	`ollama run bonsai-2:2b`
RTX 4080 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~473 tok/s	`ollama run bonsai-2:2b`
RTX 5080 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~633 tok/s	`ollama run bonsai-2:2b`
RX 7800 XT 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~411 tok/s	`ollama run bonsai-2:2b`
RX 9070 XT 16GB	Ollama	Q2_0	2 GB	fits in fast memory	~422 tok/s	`ollama run bonsai-2:2b`
M2 24GB	Ollama	Q2_0	2 GB	fits in fast memory	~88 tok/s	`ollama run bonsai-2:2b`
M3 24GB	Ollama	Q2_0	2 GB	fits in fast memory	~88 tok/s	`ollama run bonsai-2:2b`
M4 24GB	Ollama	Q2_0	2 GB	fits in fast memory	~105 tok/s	`ollama run bonsai-2:2b`
RTX 3090 24GB	Ollama	Q2_0	2 GB	fits in fast memory	~617 tok/s	`ollama run bonsai-2:2b`
RTX 4090 24GB	Ollama	Q2_0	2 GB	fits in fast memory	~665 tok/s	`ollama run bonsai-2:2b`
RX 7900 XTX 24GB	Ollama	Q2_0	2 GB	fits in fast memory	~633 tok/s	`ollama run bonsai-2:2b`
M1 Pro 32GB	Ollama	Q2_0	2 GB	fits in fast memory	~176 tok/s	`ollama run bonsai-2:2b`
M2 Pro 32GB	Ollama	Q2_0	2 GB	fits in fast memory	~176 tok/s	`ollama run bonsai-2:2b`
M4 32GB	Ollama	Q2_0	2 GB	fits in fast memory	~105 tok/s	`ollama run bonsai-2:2b`
M3 Pro 36GB	Ollama	Q2_0	2 GB	fits in fast memory	~132 tok/s	`ollama run bonsai-2:2b`
Mac Studio M4 Max (32c GPU) 36GB	Ollama	Q2_0	2 GB	fits in fast memory	~360 tok/s	`ollama run bonsai-2:2b`
Mac Studio M5 Max (32c GPU) 36GB	Ollama	Q2_0	2 GB	fits in fast memory	~404 tok/s	`ollama run bonsai-2:2b`
RTX 5090 32GB	Ollama	Q2_0	2 GB	fits in fast memory	~1182 tok/s	`ollama run bonsai-2:2b`
M4 Pro 48GB	Ollama	Q2_0	2 GB	fits in fast memory	~240 tok/s	`ollama run bonsai-2:2b`
Mac Studio M4 Max (40c GPU) 48GB	Ollama	Q2_0	2 GB	fits in fast memory	~480 tok/s	`ollama run bonsai-2:2b`
Mac Studio M5 Max (40c GPU) 48GB	Ollama	Q2_0	2 GB	fits in fast memory	~540 tok/s	`ollama run bonsai-2:2b`
M1 Max 64GB	Ollama	Q2_0	2 GB	fits in fast memory	~352 tok/s	`ollama run bonsai-2:2b`
M2 Max 64GB	Ollama	Q2_0	2 GB	fits in fast memory	~352 tok/s	`ollama run bonsai-2:2b`
M3 Max 64GB	Ollama	Q2_0	2 GB	fits in fast memory	~352 tok/s	`ollama run bonsai-2:2b`
M4 Max 64GB	Ollama	Q2_0	2 GB	fits in fast memory	~480 tok/s	`ollama run bonsai-2:2b`
Mac Studio M4 Max (40c GPU) 64GB	Ollama	Q2_0	2 GB	fits in fast memory	~480 tok/s	`ollama run bonsai-2:2b`
Mac Studio M5 Max (40c GPU) 64GB	Ollama	Q2_0	2 GB	fits in fast memory	~540 tok/s	`ollama run bonsai-2:2b`
DGX Spark 64GB (ASUS, Dell, HP, MSI, Acer, Gigabyte)	Ollama	Q2_0	2 GB	fits in fast memory	~225 tok/s	`ollama run bonsai-2:2b`
Mac Studio M3 Ultra 96GB	Ollama	Q2_0	2 GB	fits in fast memory	~720 tok/s	`ollama run bonsai-2:2b`
Mac Studio M5 Ultra 96GB	Ollama	Q2_0	2 GB	fits in fast memory	~1055 tok/s	`ollama run bonsai-2:2b`
M3 Max 128GB	Ollama	Q2_0	2 GB	fits in fast memory	~352 tok/s	`ollama run bonsai-2:2b`
M4 Max 128GB	Ollama	Q2_0	2 GB	fits in fast memory	~480 tok/s	`ollama run bonsai-2:2b`
Mac Studio M4 Max (40c GPU) 128GB	Ollama	Q2_0	2 GB	fits in fast memory	~480 tok/s	`ollama run bonsai-2:2b`
Mac Studio M5 Max (40c GPU) 128GB	Ollama	Q2_0	2 GB	fits in fast memory	~540 tok/s	`ollama run bonsai-2:2b`
DGX Spark 128GB (NVIDIA, ASUS, Dell, Lenovo, HP, MSI, Acer, Gigabyte)	Ollama	Q2_0	2 GB	fits in fast memory	~225 tok/s	`ollama run bonsai-2:2b`
M2 Ultra 192GB	Ollama	Q2_0	2 GB	fits in fast memory	~703 tok/s	`ollama run bonsai-2:2b`
Mac Studio M3 Ultra 256GB	Ollama	Q2_0	2 GB	fits in fast memory	~720 tok/s	`ollama run bonsai-2:2b`
Mac Studio M5 Ultra (64c GPU) 256GB	Ollama	Q2_0	2 GB	fits in fast memory	~1055 tok/s	`ollama run bonsai-2:2b`
Mac Studio M5 Ultra (80c GPU) 256GB	Ollama	Q2_0	2 GB	fits in fast memory	~1055 tok/s	`ollama run bonsai-2:2b`
Mac Studio M3 Ultra (80c GPU) 512GB	Ollama	Q2_0	2 GB	fits in fast memory	~720 tok/s	`ollama run bonsai-2:2b`

Estimates, not benchmarks. Model sizes are ollama.com download sizes; memory needed adds headroom for ~8k tokens of context; speeds are rough figures from memory bandwidth. Data updated 2026-10-18.

A project by OnDevice Labs · llms.txt