Bonsai 2 27B Mtp — Hardware Requirements & GPU Compatibility
Chat
Bonsai 2 27B Mtp is a 27B-parameter open language model from decent-jawfish. At Q4_K_M it needs about 17.82 GB of VRAM — see which GPUs and Macs can run it below.
Decent on a 24 GB GPU
Runs on 24 GB (Q4_K_M · ~37 tok/s · 16K)Macs from 24 GB (Q4_K_M · ~12 tok/s)
Apache 2.0
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Bonsai 2 27B Mtp
One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.
Recommended
vllm serve decent-jawfish/bonsai-2-27b-mtp --max-model-len 16384vLLM repo with its own quantisation
Download: Weights on Hugging Face
Ways to runOther ways
each opens step-by-stepRun in cloud
Runs on RTX 5090 (32 GB) (10 GB headroom) · Q4_K_M
- Generation speed
- ~65 tok/s
- generation speed
- Cost per 1M output tokens
- $2.11
- per 1M output tokens
How Much VRAM Does Bonsai 2 27B Mtp Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 12.6 GB | 20.4 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 32K context · ~52 tok/s | |
| 14.5 GB | 22.2 GB | 3-bit medium — 3-bit medium quantization | Fits: 32K context · ~45 tok/s | |
| 17.8 GB | 25.6 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 16K context · ~37 tok/s | |
| 21.2 GB | 28.9 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Fits: 8K context · ~31 tok/s | |
| 24.5 GB | 32.2 GB | 6-bit, very good — 6-bit quantization, very good quality | Does not fit: needs 28.1 GB | |
| 29.7 GB | 37.4 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Does not fit: needs 33.3 GB | |
| 59.4 GB | 67.1 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Does not fit: needs 63 GB |
Selected Q4_K_M
4-bit medium quantization21.4GBat 16K context
- Weights
- 17.0 GB
- KV cache
- 4.1 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 2.6 GB free · ~37 tok/s
≥50 tok/s needs ≥1.37 TB/s · RTX 5090 ~65
Which Hardware Can Run Bonsai 2 27B Mtp?
Q4_K_M · 17.8 GBBonsai 2 27B Mtp (Q4_K_M) requires 17.8 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 24+ GB is recommended. 8 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 122 devices with unified memory can run Bonsai 2 27B Mtp, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Mini M5 Pro (24 GB).
20 GB
tight · up to 8K ctx
1 GPU
RX 7900 XT~27Simulate output on RX 7900 XT32–40 GB
fits · 10.6 GB free · up to 32K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~65Simulate output on RTX 5090M5 Max36 GB~18Simulate output on Mac Studio M5 Max (36 GB)M532 GB~6Simulate output on MacBook Air 13" M5 (32 GB)Orin~8Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~5Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at Q4_K_M
Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Qwen3.8 27BAlibaba | 122.0 | 17.9 GB | ~38 | same size |
| Qwen3.6 27BAlibaba | 111.2 | 17.9 GB | ~38 | same size |
| Qwen3.5 27BAlibaba | 103.8 | 17.9 GB | ~38 | same size |
| Gemma 3 27B ITGoogle | 79.4 | 21.7 GB | ~36 | same size |
| Gemma 2 27B ITGoogle | 72.7 | 19.5 GB | ~37 | same size |
| Gemma 4 26B A4BGoogle | — | 16.5 GB | ~171 | same size |
Frequently Asked Questions
How much VRAM does Bonsai 2 27B Mtp need?
Bonsai 2 27B Mtp requires 17.8 GB of VRAM at Q4_K_M, or 59.4 GB at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 27B × 4.8 bits ÷ 8 = 16.2 GB
KV Cache + Overhead ≈ 1.6 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
VRAM usage by quantization
Can NVIDIA GeForce RTX 4090 run Bonsai 2 27B Mtp?
Yes, at Q5_K_M (21.2 GB) or lower. Higher quantizations like Q6_K (24.5 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.
What's the best quantization for Bonsai 2 27B Mtp?
For Bonsai 2 27B Mtp, Q4_K_M (17.8 GB) offers the best balance of quality and VRAM usage. Q5_K_M (21.2 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 12.6 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Bonsai 2 27B Mtp on a Mac?
Yes — Mac Mini M5 Pro (24 GB) and 106 other Macs can run Bonsai 2 27B Mtp. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 17.8 GB of usable unified memory (RAM minus macOS overhead).
Can I run Bonsai 2 27B Mtp locally?
Yes — Bonsai 2 27B Mtp can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Bonsai 2 27B Mtp?
At Q4_K_M, Bonsai 2 27B Mtp can reach ~269 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~37 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 17.8 × 0.65 = ~292 tok/s
Estimated speed at Q4_K_M (17.8 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Bonsai 2 27B Mtp?
At Q4_K_M, the download is about 16.20 GB. The full-precision BF16 version is 54.00 GB. The smallest option (Q2_K) is 11.47 GB.
Which GPUs can run Bonsai 2 27B Mtp?
8 consumer GPUs can run Bonsai 2 27B Mtp at Q4_K_M. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XT, AMD Radeon RX 7900 XTX. 1 GPU has plenty of headroom for comfortable inference.
Which devices can run Bonsai 2 27B Mtp?
122 devices with unified memory can run Bonsai 2 27B Mtp at Q4_K_M (17.8 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.