# Low-latency inference

Phi-mini-MoE is a lightweight Mixture of Experts (MoE) model suitable for English business and research scenarios, excelling in resource-constrained environments and low-latency scenarios.

This is a transformers model published on Hub. The specific functions and detailed information are to be supplemented.

This is a transformers model that has been uploaded to the Hugging Face Hub. Specific information is to be supplemented.

This is an automatically generated transformers model card, lacking specific model information.

This is a transformers model published on the Hugging Face Hub, and specific information is to be supplemented.

This is an INT8-quantized Mistral-Small-3.1-24B-Instruct-2503 model, optimized by Red Hat and Neural Magic, suitable for fast response and low-latency scenarios.

This is a 24B-parameter conditional generation model based on the Mistral3 architecture, optimized with FP8 dynamic quantization, suitable for multilingual text generation and visual understanding tasks.

Mistral Small 3.1 is a large multimodal language model with 24 billion parameters, possessing visual understanding ability and 128k long context processing ability, suitable for various tasks.

SANA-Sprint is an ultra-efficient text-to-image diffusion model that reduces inference steps from 20 to 1-4 while maintaining top-tier performance.

NVIDIA NeMo Canary Flash is a family of multilingual multitask models that achieves state-of-the-art performance across multiple speech benchmarks. Supports automatic speech recognition and translation tasks in four languages.

A Mistral instruction fine-tuned model with 24B parameters after INT8 quantization, significantly reducing GPU memory requirements and improving computational throughput.

Phi-4-multimodal-instruct is a lightweight open-source multimodal foundation model that integrates language, vision, and speech research and datasets from Phi-3.5 and 4.0 models. It supports text, image, and audio inputs to generate text outputs, with a context length of 128K tokens.

A fast and accurate neural machine translation model for Chinese to English translation

A distilled multilingual Whisper model supporting automatic speech recognition for 7 European languages with code-switching capability

This is the OpenVINO optimized version of the facebook/bart-large-mnli model for zero-shot text classification tasks.

A vectorization tool developed by Sinequa that generates embedding vectors from input paragraphs or queries for sentence similarity calculation and retrieval tasks.

Kotoba-Whisper is a Japanese automatic speech recognition distilled model developed by Asahi Ushio in collaboration with Kotoba Technologies, based on Whisper large-v3 distillation, achieving a 6.3x inference speed improvement.

Show-o is an any-to-any conversion model based on PyTorch, supporting input and output conversion across multiple modalities.

Zamba2-2.7B is a hybrid model composed of state space and Transformer modules, using the Mamba2 module and shared attention module, featuring high performance and low latency.

Snowflake Arctic Embed M v1.5 is an efficient sentence embedding model, focusing on sentence similarity calculation and feature extraction tasks.

MobileCLIP-B (LT) is an efficient image-text model developed by Apple, achieving fast zero-shot image classification through multimodal reinforcement training, outperforming similar models.

MobileCLIP-B is an efficient image-text model that achieves fast inference through multimodal reinforcement training and excels in zero-shot image classification tasks.

MobileCLIP-S2 is an efficient text-image model that achieves fast zero-shot image classification through multimodal reinforcement training.

MobileCLIP-S0 is an efficient image-text model achieved through multimodal reinforcement training, significantly improving speed and size efficiency while maintaining high performance.

The LLM Compiler is a state-of-the-art LLM based on Code Llama, specifically designed for code optimization and compiler inference tasks. It far exceeds existing public models in understanding compiler optimization.

Kotoba-Whisper-v1.1 is a Japanese automatic speech recognition model based on Whisper, with added punctuation and timestamp processing capabilities.

This is a Llama 3 instruction model fine-tuned for function calling, suitable for commercial use under the Llama 3 Community License.

Kotoba-Whisper is a Japanese automatic speech recognition distilled Whisper model collection jointly developed by Asahi Ushio and Kotoba Technologies, which is 6.3 times faster than the original large-v3 while maintaining similar low error rates.

Mamba is an efficient language model based on the State Space Model (SSM), with the ability to model sequences with linear time complexity.

CodeLlama 70B Python is a large language model focused on the Python programming language, based on the Llama 2 architecture, optimized for code generation and completion tasks.

CodeLlama 70B Instruct is a large-scale code generation model based on the Llama 2 architecture, specifically optimized for code understanding and generation tasks.

This is a transformers model published on the Hugging Face Hub. The specific functions and features are to be supplemented.

MobileVLM is a fast and powerful multi-modal vision-language model designed specifically for mobile devices, supporting efficient cross-modal interaction.

This is a Whisper base.en model converted based on CTranslate2, used for English speech recognition tasks.

This is the CTranslate2 converted version of the OpenAI Whisper medium.en model, used for efficient automatic speech recognition tasks.

Whisper large-v3 is a large-scale multilingual automatic speech recognition (ASR) model developed by OpenAI, supporting speech-to-text tasks in multiple languages.

A vectorizer developed by Sinequa that generates embedding vectors from input paragraphs or queries for sentence similarity computation and retrieval tasks.

A multilingual vectorizer developed by Sinequa that generates embedding vectors for input paragraphs or queries, used for similarity calculation and information retrieval.

A vectorizer developed by Sinequa capable of generating embedding vectors from paragraphs or queries for sentence similarity computation and feature extraction.

Korean automatic speech recognition model based on Conformer architecture, optimized for stream processing with excellent performance in specific domains like customer service voice