CanItRun Logocanitrun.

Bonsai 27B

Bonsai 27B doesn't use the standard quantization ladder — it ships as fixed builds needing 6.1 GB at 1-bit (Q1_0) or 9.9 GB at Ternary (Q2_0). 75 GPUs we track can run at least one build fully in VRAM at 8k context.

75 GPUs run this natively · 5 with CPU offload

PrismML27B params256k contextApache 2.0Commercial use ok

Bonsai 27B vs Qwen 3.6 27B: size & quality

Same weights as Qwen 3.6 27B, requantized to a fraction of the size.

Weights size

Qwen 3.6 27B FP16
54.0 GB
Qwen 3.6 27B Q4_K_M
15.2 GB
Bonsai 1-bit
3.8 GB
Bonsai Ternary
7.2 GB

Quality retained

Qwen 3.6 27B FP16
100% (reference)
Bonsai 1-bit
~89.5%
Bonsai Ternary
~94.6%

Quality retained is Qwen 3.6 27B's own FP16 output as the 100% reference, against PrismML's vendor-reported retention for each build — not yet verified by an independent leaderboard.

Bonsai 27B is a 27B parameter dense model developed by PrismML. Released July 2026 by PrismML as a proprietary low-bit requantization of Alibaba's Qwen3.6-27B — identical underlying weights, not a new pretrain or finetune. Two fixed builds, both Apache 2.0 (see the size chart above).

To run Bonsai 27B locally: Small enough for 8GB GPUs, base-tier Apple Silicon, or high-end phones. PrismML reports ~163 tok/s (1-bit) and ~134 tok/s (Ternary) on an RTX 5090, and ~87/58 tok/s on an M5 Max — vendor-reported, not independently verified.

Quality losses concentrate in agentic tool-calling and instruction-following rather than math (see the quality chart above). Not yet listed on any independent leaderboard.

Available builds

Bonsai 27B doesn't use the standard FP32–Q2_K quantization ladder — it's a fixed low-bit release with exactly 2 downloadable builds. Sizes below assume 8k context.

BuildBits/weightWeightsKV cacheTotalQuality vs FP16
1-bit (Q1_0)1.1253.8 GB1.61 GB6.1 GB~89.5%*
Ternary (Q2_0)1.587.2 GB1.61 GB9.9 GB~94.6%*

*Vendor-reported on their own benchmark suite — not yet verified by an independent leaderboard. KV cache shown at 8k context (FP16).

Benchmarks

Independent benchmark scores for Bonsai 27B are not yet listed on the Open LLM Leaderboard v2. Third-party requantization of Qwen3.6-27B by PrismML — same weights, not a retrain. Doesn't use the standard FP32-Q2_K ladder shown for other models; it ships only as two fixed sub-2-bit builds (see the quantization table on this page). PrismML's own benchmark suite reports 89.5%–94.6% of FP16 quality retained across the two builds; not yet independently verified by a third party. See the HuggingFace model card for reported evaluations.

GPUs that run Bonsai 27B natively (75)

Plus 5 GPUs that run it with CPU offload (slower)

Notes

Third-party requantization of Qwen3.6-27B by PrismML — same weights, not a retrain. Doesn't use the standard FP32-Q2_K ladder shown for other models; it ships only as two fixed sub-2-bit builds (see the quantization table on this page). PrismML's own benchmark suite reports 89.5%–94.6% of FP16 quality retained across the two builds; not yet independently verified by a third party.

Hugging Face ↗Released 2026-07-14

GPUs mentioned

NVIDIA RTX 509032 GB
NVIDIAConsumer1792 GB/s
NVIDIA RTX 508016 GB
NVIDIAConsumer960 GB/s
NVIDIA RTX 5070 Ti16 GB
NVIDIAConsumer896 GB/s
NVIDIA RTX 507012 GB
NVIDIAConsumer672 GB/s
NVIDIA RTX 5060 Ti 16GB16 GB
NVIDIAConsumer448 GB/s
NVIDIA RTX 50608 GB
NVIDIAConsumer448 GB/s
NVIDIA RTX 50508 GB
NVIDIAConsumer320 GB/s
NVIDIA RTX 409024 GB
NVIDIAConsumer1008 GB/s
NVIDIA RTX 408016 GB
NVIDIAConsumer717 GB/s
NVIDIA RTX 4070 Ti12 GB
NVIDIAConsumer504 GB/s
NVIDIA RTX 407012 GB
NVIDIAConsumer504 GB/s
NVIDIA RTX 4060 Ti 16GB16 GB
NVIDIAConsumer288 GB/s
NVIDIA RTX 40608 GB
NVIDIAConsumer272 GB/s
NVIDIA RTX 309024 GB
NVIDIAConsumer936 GB/s
NVIDIA RTX 3090 Ti24 GB
NVIDIAConsumer1008 GB/s
NVIDIA RTX 3080 10GB10 GB
NVIDIAConsumer760 GB/s
NVIDIA RTX 3060 12GB12 GB
NVIDIAConsumer360 GB/s
NVIDIA H100 80GB80 GB
NVIDIADatacenter3350 GB/s
NVIDIA A100 80GB80 GB
NVIDIADatacenter2039 GB/s
NVIDIA A100 40GB40 GB
NVIDIADatacenter1555 GB/s
NVIDIA L40S48 GB
NVIDIADatacenter864 GB/s
NVIDIA RTX A600048 GB
NVIDIAWorkstation768 GB/s
NVIDIA RTX 4000 Ada20 GB
NVIDIAWorkstation320 GB/s
NVIDIA RTX 4500 Ada24 GB
NVIDIAWorkstation432 GB/s
NVIDIA RTX 5000 Ada32 GB
NVIDIAWorkstation576 GB/s
NVIDIA RTX 6000 Ada48 GB
NVIDIAWorkstation960 GB/s
NVIDIA RTX Pro 600096 GB
NVIDIAWorkstation1344 GB/s
NVIDIA DGX Spark (128GB)128 GB
NVIDIAWorkstation273 GB/s
AMD Radeon RX 7900 XTX24 GB
AMDConsumer960 GB/s
AMD Radeon RX 7900 XT20 GB
AMDConsumer800 GB/s
AMD Radeon RX 7900 GRE16 GB
AMDConsumer576 GB/s
AMD Radeon RX 6800 XT16 GB
AMDConsumer512 GB/s
AMD Radeon PRO W780032 GB
AMDWorkstation576 GB/s
AMD Radeon PRO W790048 GB
AMDWorkstation864 GB/s
AMD Instinct MI300X192 GB
AMDDatacenter5300 GB/s
AMD Radeon AI Pro 9700 32GB32 GB
AMDDatacenter640 GB/s
AMD Strix Halo (128GB)128 GB
AMDLaptop256 GB/s
AMD Strix Halo (96GB)96 GB
AMDLaptop256 GB/s
AMD Strix Halo (64GB)64 GB
AMDLaptop256 GB/s
Apple M5 Max (128GB)128 GB
AppleLaptop614 GB/s
Apple M5 Max (64GB)64 GB
AppleLaptop614 GB/s
Apple M5 Max (48GB)48 GB
AppleLaptop614 GB/s
Apple M5 Pro (48GB)48 GB
AppleLaptop307 GB/s
Apple M4 Ultra (384GB)384 GB
AppleWorkstation1092 GB/s
Apple M4 Ultra (192GB)192 GB
AppleWorkstation1092 GB/s
Apple M4 Max (128GB)128 GB
AppleLaptop546 GB/s
Apple M4 Max (96GB)96 GB
AppleLaptop546 GB/s
Apple M4 Max (64GB)64 GB
AppleLaptop546 GB/s
Apple M4 Max (48GB)48 GB
AppleLaptop546 GB/s
Apple M4 Pro (48GB)48 GB
AppleLaptop273 GB/s
Apple M3 Ultra (512GB)512 GB
AppleWorkstation819 GB/s
Apple M3 Ultra (256GB)256 GB
AppleWorkstation819 GB/s
Apple M3 Ultra (96GB)96 GB
AppleWorkstation819 GB/s
Apple M3 Max (128GB)128 GB
AppleLaptop400 GB/s
Apple M3 Max (96GB)96 GB
AppleLaptop400 GB/s
Apple M3 Max (64GB)64 GB
AppleLaptop400 GB/s
Apple M3 Max (48GB)48 GB
AppleLaptop400 GB/s
Apple M2 Ultra (384GB)384 GB
AppleWorkstation800 GB/s
Apple M2 Ultra (192GB)192 GB
AppleWorkstation800 GB/s
Apple M2 Max (96GB)96 GB
AppleLaptop400 GB/s
Apple M2 Max (64GB)64 GB
AppleLaptop400 GB/s
Apple M1 Ultra (128GB)128 GB
AppleWorkstation800 GB/s
Apple M1 Ultra (64GB)64 GB
AppleWorkstation800 GB/s
Apple M1 Max (64GB)64 GB
AppleLaptop400 GB/s
Intel Arc B580 12GB12 GB
IntelConsumer456 GB/s
Intel Arc B570 10GB10 GB
IntelConsumer380 GB/s
Intel Arc Pro B70 24GB24 GB
IntelWorkstation456 GB/s
Intel Arc Pro B60 24GB24 GB
IntelWorkstation380 GB/s
Intel Arc A770 16GB16 GB
IntelConsumer560 GB/s
Intel Arc A770 8GB8 GB
IntelConsumer512 GB/s
Intel Arc A750 8GB8 GB
IntelConsumer512 GB/s
Intel Arc A580 8GB8 GB
IntelConsumer512 GB/s
Intel Arc Pro A60 12GB12 GB
IntelWorkstation384 GB/s
Intel Data Center GPU Max 1550128 GB
IntelDatacenter3276 GB/s
Intel Data Center GPU Max 110048 GB
IntelDatacenter1229 GB/s

Frequently asked questions

What are the VRAM requirements for Bonsai 27B?
Bonsai 27B doesn't use the standard quantization ladder — it ships as 2 fixed builds: 1-bit (Q1_0) (6.1 GB) and Ternary (Q2_0) (9.9 GB). These figures assume a 8k context window; VRAM scales linearly with context length due to the KV cache.
How many parameters does Bonsai 27B have?
Bonsai 27B has 27 billion parameters — the same count as the base model it's derived from, since this release is a post-training requantization rather than a retrain.
What quantization levels does Bonsai 27B come in?
Bonsai 27B skips the usual FP16-to-Q2_K ladder entirely. It's only available as: 1-bit (Q1_0) at 1.125 bits/weight (6.1 GB total, ~89.5% of FP16 quality per the vendor); Ternary (Q2_0) at 1.58 bits/weight (9.9 GB total, ~94.6% of FP16 quality per the vendor).
Can Bonsai 27B run on a small GPU?
Yes. The 1-bit (Q1_0) build needs only 6.1 GB, which fits on most 8 GB+ GPUs, entry-level Apple Silicon, and even high-end phones per the vendor's own claims.
What GPU do I need to run Bonsai 27B locally?
Either build is small enough for almost any GPU released in the last several years. The Ternary (Q2_0) build (9.9 GB) needs a bit more headroom for better quality, but both fit comfortably on an 8 GB card or larger.