PrismML27B1-bit (binary, ~1.125 bpw) 下约 5 GB 显存

1-bit Bonsai 27B — 显存、速度与本地部署

作者: Jakub Rusinowski · 最后更新:

The phone build. Binary {−1, +1} weights at ~1.125 bits/weight bring a full 27B multimodal model down to 3.9 GB — small enough to run on an iPhone 17 Pro at ~11 tok/s. PrismML reports it keeps more than 90% of the FP16 baseline (MATH 91.66, coding 81.88); tool-calling is the weak spot, dropping from ~80 to ~66. Ships as GGUF (llama.cpp / LM Studio) and MLX (Apple). Specs from launch coverage — verify on the Hugging Face model card.

1-bit Bonsai 27B 在 1-bit (binary, ~1.125 bpw) 下约需 5 GB 显存——量化权重加框架开销,不含 KV 缓存。在 Apple Silicon 上,这部分来自统一内存。

逻辑87
创意80
编程85

按量化级别的显存与速度

计算基准:NVIDIA RTX 4090 (24 GB)。仅含权重与开销:该模型架构未公开,因此未计入 KV 缓存。

量化显存速度(估算)适配
1-bit (binary, ~1.125 bpw)
1.13 bpw
4.6 GB~119 tok/s可运行

黑色标记 = NVIDIA RTX 4090 (24 GB) 上的可用显存。 估算来自内存带宽屋顶线模型,详见 方法说明页. 1-bit Bonsai 27B 显存计算器 →

运行 1-bit Bonsai 27B

目录中能运行 1-bit Bonsai 27B 的最便宜 GPU 是 Intel Arc B570 (10 GB).

购买此硬件 Intel Arc B570 10GB — 10 GB VRAM · 150 W board power立即云端部署 RunPod 上的 RTX 4090 — 低至 $0.34/小时 · 价格核实于 2026-07

或在 Vast.ai 比较,低至 $0.35/小时 (typical low · varies)

作为亚马逊联盟成员,我们从符合条件的购买中获得收入。云 GPU 链接为推荐链接——我们可能获得佣金,您无需额外付费。

联盟营销声明: 本页部分链接为联盟推广链接——如果你通过它们购买,LLM Configurator 可能会获得佣金,而你无需支付任何额外费用。作为亚马逊联盟成员(Amazon Associate),LLM Configurator 会从符合条件的购买中获得收益。
Intel Arc B570 10GB
10 GB VRAM · 150 W board power
2026年价格波动较大——请以当前商品页价格为准。

如何运行 1-bit Bonsai 27B

安装 Ollama,然后运行:

ollama run bonsai-27b
Hugging Face 上的权重: prism-ml/Bonsai-27B-gguf ↗

规格

Preview — The model is released, but these specs are thin or rest on a single source. Individual fields may be wrong.

Preview — The model is released, but these specs are thin or rest on a single source. Individual fields may be wrong.

参数量
27 Billion
上下文窗口
262,144
架构
Qwen3.6-27B (multimodal), 1-bit binary weights
提供商
PrismML
许可证
Apache 2.0
Practical deployment
Edge device (phone, SBC)
规格量化
1-bit (binary, ~1.125 bpw)
系统内存
8 GB
记录更新于
2026-07-15
许可证Apache-2.0允许商业使用

Commercial use permitted. No usage restrictions beyond attribution.

质量与使用场景

评分由模型作者或独立评测方发布——衡量质量而非吞吐量,并非我们实测。

最适合on devicephoneofflinemultimodalreasoning
基准分数来源
MATH91.66 %厂商声称 · PrismML (launch)
Coding (HumanEval+ class)81.88 %厂商声称 · PrismML (launch)
Tool-calling (BFCL class)66 %厂商声称 · PrismML (launch)

我的 GPU 能运行 1-bit Bonsai 27B 吗?

Bonsai 27B 的其他尺寸

1-bit Bonsai 27B — 常见问题

How much VRAM does 1-bit Bonsai 27B need?

About 5 GB at 1-bit (binary, ~1.125 bpw) — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.

Does 1-bit Bonsai 27B run on an RTX 4090 (24 GB)?

Yes. 1-bit Bonsai 27B needs about 5 GB at 1-bit (binary, ~1.125 bpw), inside a 24 GB card, at an estimated 119 tokens/sec.

How do I run 1-bit Bonsai 27B locally?

Install Ollama and run `ollama run bonsai-27b`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.

What other sizes does Bonsai 27B come in?

1-bit Bonsai 27B (5 GB), Ternary Bonsai 27B (7 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.