GLM Alternatives

This repository hosts an AWQ 4-bit quantized version of the GLM-4.7-Flash model, optimized for reduced memory usage and faster inference. It retains tool-calling capabilities and a full chat template, making it suitable for local or hosted deployment where full-precision models would be too large. The model is provided for developers seeking efficient open-weight alternatives.

GLM-5.2-Int4-Int8Mix is a quantized variant of the GLM-5.2 model using a mix of 4-bit and 8-bit precision. It includes support for reasoning effort levels, tool calling, and advanced system prompting. The model is designed for efficient local or server-based inference while maintaining strong language understanding and generation capabilities.

GLM-5.2-AWQ-INT4 is an INT4 quantized variant of the GLM-5.2 model optimized for efficient inference. It includes support for adjustable reasoning effort levels and function calling. The model is available on Hugging Face and is designed for local deployment using quantization-aware runtimes.

GLM-4.5 is an open-source large language model supporting vision inputs, tool calling, and advanced chat templates. It can process images and text together and follows structured function-calling formats. The model is distributed on Hugging Face and supports Docker and pip-based inference.

unsloth/GLM-4.7-Flash-GGUF contains GGUF quantized files for the GLM-4.7 Flash model. It includes support for tool calling using XML-style function call formatting. The model is suitable for local deployment using llama.cpp or other GGUF-compatible runtimes and is provided to enable efficient on-premise or desktop AI applications.

GLM-4.5-Air-FP8 is an FP8 quantized variant of the GLM-4.5 model, designed for efficient local inference. It supports tool calling and includes a comprehensive chat template. The model is suitable for developers building local AI applications and is distributed via Hugging Face for use with standard inference frameworks.

This repository contains an NVFP4 quantized variant of GLM-5.2 for optimized inference on NVIDIA hardware. It includes support for tool calling and advanced reasoning modes. The model follows a specific system prompt format and can be used with compatible inference engines.

GLM-4.7-Flash-NVFP4 is a quantized variant of the GLM-4.7 language model hosted on Hugging Face. It is provided by user GadflyII under the repository name GLM-4.7-Flash-NVFP4 and uses NVFP4 precision for reduced memory and compute requirements during inference. The model includes a chat template that supports tool calling. When tools are supplied, the template instructs the model to output function calls in a specific XML format enclosed in tool_call tags, with individual arguments expressed as arg_key and arg_value pairs. A visible_text macro handles content rendering for strings, iterables, or mappings containing text items. The tokenizer configuration specifies an end-of-text token as the pad token. It is distributed as a model repository on the Hugging Face platform. Users obtain the files through the standard Hugging Face ecosystem for loading with compatible inference libraries. The page provides no information on licensing, pricing, or intended audience beyond the general Hugging Face context of open-source and open-science AI advancement.

This repository provides GGUF quantized weights of the GLM-4.7-Flash model, optimized for use with LM Studio, llama.cpp, and other GGUF-compatible engines. It includes support for tool calling and is intended for local CPU/GPU inference of the GLM series of large language models.

An AMD-published quantized version of the GLM-5.2 large language model using MXFP4 precision. The repository includes chat templates supporting tool use, reasoning effort levels, and system prompts. It is designed for compatibility with Hugging Face pipelines and AMD inference runtimes.

GLM-4.7-Flash is a foundation model hosted on Hugging Face under the repository zai-org/GLM-4.7-Flash. It belongs to the class of large language models that process text through a defined chat template. The model includes a chat template in Jinja format that supports tool use. When tools are supplied, the template instructs the model to consider function signatures provided inside XML-style tags and to output calls in a specific XML format containing the function name along with argument keys and values. A visible_text macro within the template handles string content, iterable collections, and mapping objects by extracting text where present. The tokenizer configuration specifies an end-of-text token as the pad token. The model is delivered as a repository on the Hugging Face platform, where users can access the associated configuration files. Hugging Face itself operates to advance and democratize artificial intelligence through open source and open science. No pricing, licensing terms, or specific target audience beyond the general context of model repositories are stated.

GLM-4.5-Air is an efficient variant of the GLM-4.5 series of large language models. It supports tool calling and follows a chat template suitable for instruction and agentic use cases. The model is hosted on Hugging Face and intended for inference with the Transformers library or compatible frameworks.

This is an AWQ-quantized 4B parameter version of the Qwen 3.5 model optimized for efficient inference. It supports advanced features including tool use and vision capabilities. The model is distributed on Hugging Face for use with Transformers or local inference engines.

Qwen3.6-27B-AWQ-6Bit is a quantized version of the Qwen3.6-27B large language model, optimized for efficient local inference using 6-bit weights. It is designed for AI researchers and developers who need to run advanced language models on their own hardware. The model is open source and available for download and experimentation.

GLM-5.2 is an open-source large language model designed for advanced text generation and natural language processing tasks. It is suitable for developers and researchers seeking customizable, local deployment of LLMs with open weights.

A GGUF quantized version of the GLM-5.2 model provided by Unsloth. It supports advanced features such as tool calling and high-reasoning modes. The model is designed for efficient local deployment using inference engines that support the GGUF format, making powerful language models accessible on standard hardware.

This is an AWQ-quantized 27B parameter version of Alibaba's Qwen3.5 language model. It supports text generation, tool use, and multimodal inputs while requiring significantly less memory than the original. The model is distributed via Hugging Face for use with popular inference frameworks.

GLM-4.7-Flash-MLX-8bit is a community-quantized version of the GLM-4 large language model using 8-bit precision and optimized for the MLX framework on Apple silicon. It supports tool calling and can be used locally through libraries such as Transformers or MLX. The model is hosted on Hugging Face for easy download and integration into local AI applications.

GLM-4.5V is a vision-enabled version of the GLM-4.5 large language model. It supports image inputs, tool calling via structured XML formats, and custom chat templates. The model weights are hosted on Hugging Face for developers to integrate into applications or run locally.

GLM-4.7-Flash is an optimized, high-speed variant of Zhipu AI's GLM-4 model, prepared by Unsloth for efficient deployment. It supports tool calling, structured output, and standard chat templates. The model is aimed at developers who require fast, cost-effective language model inference for production applications.

This is a 6-bit quantized version of the GLM-4.7-Flash model, prepared by the LM Studio community for efficient inference using the MLX framework on Apple devices. It includes support for tool calling and follows a standard chat template. The model is hosted on Hugging Face for easy local deployment.