Ternary Bonsai 2 系列 AI 模型登场 苹果 iPhone 18 Pro Max 上本地吞吐最高每秒 120 个词元Ternary Bonsai 2 Series AI model debuts on Apple iPhone 18 Pro Max with native throughput up to 120 words per second


AI free token/AI免费积分

IT之家 10 月 16 日消息,AI 公司 PrismML 昨日(10 月 15 日)发布博文,宣布推出 Ternary Bonsai 2 系列模型,共有 2B、4B 和 9B 三种规模,目标在手机、PC 以及可穿戴设备上本地运行 AI 智能体,执行推理、识图、调用工具等能力。

Ternary Bonsai 2 系列 AI 模型登场 苹果 iPhone 18 Pro Max 上本地吞吐最高每秒 120 个词元

在能力上,Bonsai 2 系列强调本地智能体的实用性:遵循指令、完成日常任务、在移动设备上调用工具、避免重复输出等。官方在 19 项基准上测试了智能工具调用、数学、指令遵循、编程、知识与推理、视觉理解六类能力;其中 9B 模型在标准指令执行和日常数学任务中,所需 token 数量约减少 40%。

表现维度,PrismML 表示,在 RTX 4090 和 M5 Pro 等硬件上,新模型相较对应全精度模型可达成最高 5 倍吞吐量和 3 倍能效;在 iPhone 18 Pro Max 上,9B 模型最高可达每秒 34 个词元,2B 模型最高可达每秒 120 个词元。IT之家附上相关截图如下:

Ternary Bonsai 2 系列 AI 模型登场 苹果 iPhone 18 Pro Max 上本地吞吐最高每秒 120 个词元

该系列可与已有的 Bonsai 2 27B 组成从轻量任务到复杂工作流的完整机型线。开发者可借助 llama.cpp 在 NVIDIA GPU 上运行,也可借助 MLX 在 Mac、iPhone 和 iPad 上部署;模型权重以 Apache 2.0 许可证开放。

本文来源:https://www.ithome.com/1/013/894.htm
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。

IT House October 16th news, AI company PrismML yesterday (October 15th) published a blog post, announced the launch of the Ternary Bonsai 2 series model, a total of 2B, 4B and 9B three sizes, the goal is to run AI agents locally on mobile phones, PCs and wearable devices, perform inference, map, call tools and other capabilities.

Ternary Bonsai 2 Series AI model debuts on Apple iPhone 18 Pro Max with native throughput up to 120 words per second

In terms of capabilities, the Bonsai 2 series emphasizes the usefulness of local agents: following instructions, completing daily tasks, invoking tools on mobile devices, avoiding duplicate outputs, etc. Six types of abilities were tested on 19 benchmarks: smart tool call, mathematics, instruction compliance, programming, knowledge and reasoning, and visual comprehension; of these, the 9B model reduced the number of tokens required in standard instruction execution and daily mathematical tasks by about 40%.

Expression dimension, PrismML said that on hardware such as RTX 4090 and M5 Pro, the new model can achieve up to 5 times throughput and 3 times energy efficiency compared to the corresponding full-precision model; on the iPhone 18 Pro Max, the 9B model can reach up to 34 words per second, and the 2B model can reach up to 120 words per second. IT Home is accompanied by the following screenshots:

Ternary Bonsai 2 Series AI model debuts on Apple iPhone 18 Pro Max with native throughput up to 120 words per second

The series can be combined with the existing Bonsai 2 27b to form a complete model line from lightweight tasks to complex workflows. Developers can run on NVIDIA GPUs with llama.cpp or deploy on Macs, iPhones, and iPads with MLX; model weights are available under the Apache 2.0 license.

This article is from: https://www.ithome.com/1/013/894.htm For copyright issues, please write to: 13951431913#139.com or submit a review on the right.

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry