Bob Documentation
website·Indexed 34 days ago
https://bobtranslate.com/Documentation for Bob, featuring guides on integrating translation, text-to-speech (TTS), and OCR services. It includes setup instructions for various AI providers such as OpenAI, Claude, DeepL, Google, and Microsoft, as well as local model runners like Ollama and LM Studio, and developer documentation for creating plugins using JavaScript.
What's inside Bob
- SiliconFlow provides SiliconCloud, a one-stop large model cloud service platform utilizing the SiliconLLM inference engine and OneDiff for high-performance image/video generation. It offers a wide range of LLMs, embedding/re-ranking models, and multimodal models. Certain open-source models such as Qwen2 (7B), GLM4 (9B), and Yi1.5 (9B) are available for long-term free use.
Overview of LM Studio
LM Studio is a graphical user interface (GUI) application that allows users to run large language models (LLMs) locally on their computers, similar to Ollama. Running models locally consumes system resources, and performance depends on the chosen model and the computer's hardware specifications. Using LM Studio for translation is offline and free.Overview of AIHubMix API Routing Service
AIHubMix provides an AI model API routing service based on the unified OpenAI API standard. It supports all official OpenAI models as well as other mainstream large models such as Claude, Gemini, DeepSeek, and Alibaba Qwen. The service is operated by AIHubMix, LLC (USA) and utilizes official quotas from Microsoft Azure, AWS, and GCP.Overview of the $data binary data class
The$dataclass is a wrapper aroundNSData(Objective-C) provided for the plugin environment. It is primarily used for handling binary data, such as image data required by text recognition plugins.Overview of Bob translation features
Bob provides a translation toolset that supports three primary interaction methods: selection-based translation (highlighting text), screenshot translation (OCR), and manual input translation. It allows users to enable multiple translation services simultaneously and includes support for the Concise English-Chinese Dictionary.Overview of Bob Plugin JavaScript APIs
Bob plugins are developed using JavaScript. To extend functionality, Bob provides a set of built-in APIs. With the exception ofrequire, all Bob-provided APIs start with the$symbol to distinguish them from standard JavaScript objects.Doubao model pricing overview
Pricing for common Doubao models is based on tokens per million (1M tokens). Note that pricing is subject to change and should be verified via official Volcengine documentation.Model Input (per 1M tokens) Output (per 1M tokens) Doubao-pro-32k 0.8 CNY 2 CNY Doubao-lite-32k 0.3 CNY 0.6 CNY Use Zhipu GLM models in Bob Translate
Bob Translate integrates with Zhipu GLM models. The GLM-4-Flash model is provided for free to all Bob users and does not require an API Key. For higher translation quality, users can opt for the GLM-4-Plus flagship model, which requires a self-applied API Key.Extend Bob services via plugins
Bob allows users to implement custom APIs by developing and installing plugins. Once a plugin is installed, custom services can be used just like built-in services. This applies to both general translation services and OCR services.Restore Option+Letter shortcuts on macOS 15 with BobHelper
BobHelper is a standalone companion app for Bob that restores the ability to use 'Option + Letter' shortcuts on macOS 15 Sequoia, which were previously restricted by the system.
Compatibility Requirements:
- Bob version must be 1.11.0 or higher.
- Only required for users on macOS 15.0, 15.1, and 15.2 Beta 1.
- Users on macOS 15.2 (Build 24C5073e) or later do not need BobHelper as Apple fixed the shortcut issue in the OS.
Conflict Warning: If you are using Alfred Workflows or Raycast Extensions to solve this shortcut issue, disable them before using BobHelper.
Understand Bob's service architecture for Translation and OCR
Bob relies on specific services to power its core functional modules. Each module is supported by one or more services:
- Translation Module: Uses Text Translation (called every time a translation occurs), Text Recognition (called during screenshot translation to extract text), and Speech Synthesis (called when clicking the play button on the translation page).
- OCR Module: Uses only the Text Recognition service.
Configure ERNIE (文心一言) translation service in Bob
To use ERNIE (文心一言) for translation in Bob, follow these steps:
- Register and log in to the Baidu AI Cloud Qianfan Platform.
- Navigate to 'System Management and Statistics' (系统管理与统计) > 'API Key' to obtain your API Key.
- In Bob, go to 'Preferences' (偏好设置) > 'Services' (服务).
- Select 'Text Translation' (文本翻译), click the
+button, select 'ERNIE' (文心一言), and enter your API Key.
Note: As of Bob version 1.15.0 (updated 2025/04/13), the service has been upgraded to API V2. Users of older versions must re-acquire their API keys and re-add the service.