Home/Inference & Serving/MOSS

MOSS

OpenMOSS/MOSS

An open-source conversational language model

GitHub Website

GraphCanon updated 5d · GitHub synced 5d · 31 views this month

12k stars1.1k forksLast push 2mo Python Apache-2.0

Decision brief

An open-source conversational language model from Fudan University providing pre-trained and fine-tuned models for various applications.

Good fit when

- MOSS is ideal for use in scenarios that require detailed multi-turn dialogues with advanced plugin capabilities, such as customer support services where context preservation and the ability to call
- Use MOSS when you need a model that can handle both general text generation tasks and specialized conversational settings demanding interaction with external plugins like search engines or calculers

Avoid when

- Avoid using MOSS in situations where you require models without integrated plugin support, as its advanced feature set might introduce unnecessary complexity.
- MOSS may not be the optimal choice for applications that prioritize extremely low resource consumption because of its demand for significant computational power even with the lower quantized models.

Requirements:: Min 16 GB RAM; Requires substantial GPU memory, ranging from around 12GB to 24GB depending on the model version.; Hardware must support high-performance matrix operations for effective inference.

Observed Jul 11, 2026 · Source: enrich:decision_facts

MOSS alternatives Trust report

MOSS vs jan MOSS vs worldmonitor

Verify the decision

Maintenance and security

Full trust report

Maintenance: Steady (81d since push); As of 5d
Provenance: Not a fork · Organization account; As of 5d
Security (OSV): 73 low (73 low); As of 1mo

Public GitHub metadata and optional OSV scans. Signals, not a guarantee. Trust methodology.

Install

pip install MOSS

PyPI

How it fits your stack(9)

Typed graph edges - alternatives, integrations, successors, and dependencies. Ranked by relationship type, not raw GitHub stars.

Alternative

worldmonitor

MOSS is an open-source conversational language model that supports plugin usage and various quantization levels for computational efficiency. Worldmonitor is an AI-powered platform for real-time global news aggregation, geopolitical monitoring, and infrastructure tracking. The 'alternative' relationship between MOSS and worldmonitor stems from their divergent focus areas; while MOSS offers a tool-

alternative

jan

Both MOSS and Jan are open-source conversational language models designed to interact with users in a natural way, addressing similar needs but developed independently.

alternative

Integrates

transformers

MOSS integrates with transformers because it leverages the framework's extensive library of pre-trained models and tools for developing its conversational capabilities, facilitating both training and inference processes.

integrates

langchain

MOSS, as an advanced conversational model, can be integrated into LangChain to enhance its capabilities in handling dialogues and generating human-like responses.

integrates

unsloth

MOSS can integrate with Unsloth to train and run open models locally, leveraging Unsloth's web UI for easy model handling.

integrates

llm-course

The MOSS project can be a practical application complement to the llm-course which provides educational resources on large language models.

gpt_academic

Both focus on enhancing the interactivity and utility of large language models (LLMs) in specific contexts, with Binary Husky GPT Academic emphasizing academic use cases such as paper reading and writing.

Relationship graph

Optional deeper exploration of typed edges and category neighbours.

Langchain-Chatchat39k★

llm-course82k★

gpt_academic71k★

AlternativeIntegratesRelated

Langchain-Chatchat39k★

llm-course82k★

gpt_academic71k★

AlternativeIntegratesRelated

Similar tools

Same-category neighbours not already linked as typed edges.

tensorflow

An Open Source Machine Learning Framework for Everyone

197k★

AutoGPT

AutoGPT is the vision of accessible AI for everyone, to use and to build on.

187k★

ollama

Get up and running with various large language models using Ollama.

178k★

Evidence and technical details

Sourced facts, taxonomy, compatibility claims, README excerpt, and machine-readable endpoints.

Overview

A tool-augmented conversational language model developed by Fudan University, offering various pre-trained and fine-tuned models for different use cases including general purpose text generation and plugin-enhanced multi-turn dialogues.

Capability facts

Languages: python
Source: github.language · Aug 17, 2026

Graph entities

Smolagentsframework

README

MOSS

[论文][中文版] [English] [官方微信群]

:spiral_notepad: 开源清单

模型

moss-moon-003-base: MOSS-003基座模型，在高质量中英文语料上自监督预训练得到，预训练语料包含约700B单词，计算量约6.67x10²²次浮点数运算。
moss-moon-003-sft: 基座模型在约110万多轮对话数据上微调得到，具有指令遵循能力、多轮对话能力、规避有害请求能力。
moss-moon-003-sft-plugin: 基座模型在约110万多轮对话数据和约30万插件增强的多轮对话数据上微调得到，在moss-moon-003-sft基础上还具备使用搜索引擎、文生图、计算器、解方程等四种插件的能力。
moss-moon-003-sft-int4: 4bit量化版本的moss-moon-003-sft模型，约占用12GB显存即可进行推理。
moss-moon-003-sft-int8: 8bit量化版本的moss-moon-003-sft模型，约占用24GB显存即可进行推理。
moss-moon-003-sft-plugin-int4: 4bit量化版本的moss-moon-003-sft-plugin模型，约占用12GB显存即可进行推理。
moss-moon-003-sft-plugin-int8: 8bit量化版本的moss-moon-003-sft-plugin模型，约占用24GB显存即可进行推理。
moss-moon-003-pm: 在基于moss-moon-003-sft收集到的偏好反馈数据上训练得到的偏好模型，将在近期开源。
moss-moon-003: 在moss-moon-003-sft基础上经过偏好模型moss-moon-003-pm训练得到的最终模型，具备更好的事实性和安全性以及更稳定的回复质量，将在近期开源。
moss-moon-003-plugin: 在moss-moon-003-sft-plugin基础上经过偏好模型moss-moon-003-pm训练得到的最终模型，具备更强的意图理解能力和插件使用能力，将在近期开源。

数据

moss-002-sft-data: MOSS-002所使用的多轮对话数据，覆盖有用性、忠实性、无害性三个层面，包含由text-davinci-003生成的约57万条英文对话和59万条中文对话。
moss-003-sft-data: moss-moon-003-sft所使用的多轮对话数据，基于MOSS-002内测阶段采集的约10万用户输入数据和gpt-3.5-turbo构造而成，相比moss-002-sft-data，moss-003-sft-data更加符合真实用户意图分布，包含更细粒度的有用性类别标记、更广泛的无害性数据和更长对话轮数，约含110万条对话数据。完整数据已全部开源。
moss-003-sft-plugin-data: moss-moon-003-sft-plugin所使用的插件增强的多轮对话数据，包含支持搜索引擎、文生图、计算器、解方程等四个插件在内的约30万条多轮对话数据。已开源所有数据。
moss-003-pm-data: moss-moon-003-pm所使用的偏好数据，包含在约18万额外对话上下文数据及使用moss-moon-003-sft所产生的回复数据上构造得到的偏好对比数据，将在近期开源。

工程方案

MOSS Vortex - MOSS部署和推理方案
MOSS WebSearchTool - MOSS搜索引擎插件部署方案
MOSS Frontend - 基于flutter实现的MOSS-003前端界面
MOSS Backend - 基于Go实现的MOSS-003后端

:fountain_pen: 介绍

MOSS是一个支持中英双语和多种插件的开源对话语言模型，moss-moon系列模型具有160亿参数，在FP16精度下可在单张A100/A800或两张3090显卡运行，在INT4/8精度下可在单张3090显卡运行。MOSS基座语言模型在约七千亿中英文以及代码单词上预训练得到，后续经过对话指令微调、插件增强学习和人类偏好训练具备多轮对话能力及使用多种插件的能力。

局限性：由于模型参数量较小和自回归生成范式，MOSS仍然可能生成包含事实性错误的误导性回复或包含偏见/歧视的有害内容，请谨慎鉴别和使用MOSS生成的内容，请勿将MOSS生成的有害内容传播至互联网。若产生不良后果，由传播者自负。

MOSS用例：

简单数学应用题

解方程

生

For agents

This page has a .md twin and JSON over the API.

/tools/openmoss-moss.md GET /api/graphcanon/tools/openmoss-moss GET /api/graphcanon/graph?tool=openmoss-moss

MOSS

Decision brief

Good fit when

Avoid when

Verify the decision

Maintenance and security

Install

How it fits your stack(9)

Alternative

Integrates

Related

Relationship graph

Similar tools

Evidence and technical details

Overview

Capability facts

Categories

Graph entities

Tags

README

MOSS

目录

:spiral_notepad: 开源清单

模型

数据

工程方案

:fountain_pen: 介绍

For agents