# Low-Parameter Efficiency

Qwen3-0.6B is the latest 0.6B-parameter large language model in the Qwen series, supporting mode switching between reasoning and non-reasoning modes, with powerful reasoning, instruction-following, and multilingual capabilities.

Qwen3 is the latest generation of large language models in the Tongyi Qianwen series, offering both dense models and Mixture-of-Experts (MoE) architectures. Trained on massive datasets, Qwen3 achieves breakthrough progress in reasoning, instruction following, agent capabilities, and multilingual support.

The latest version of the Tongyi Qianwen series of large language models, supporting switching between thinking and non-thinking modes, with powerful reasoning, multilingual, and agent capabilities.

Falcon-H1 is a causal decoder-only language model with a hybrid Transformers+Mamba architecture developed by TII, supporting English with outstanding performance.

Qwen3-1.7B is the 1.7B parameter version of the latest generation in the Qwen series of large language models, supporting mode switching, multilingual processing, and agent capabilities.

OpenMath-Nemotron-1.5B is a mathematical reasoning model fine-tuned on the OpenMathReasoning dataset based on Qwen2.5-Math-1.5B, achieving state-of-the-art results on multiple mathematical benchmarks.

UI-R1 is a vision-language model enhanced by reinforcement learning for GUI agent action prediction, built upon Qwen2.5-VL-3B-Instruct.

DRAMA-base is a dense retrieval model based on a pruned large language model architecture, supporting multilingual text retrieval tasks.

TIGER is a lightweight speech separation model that effectively extracts key acoustic features through frequency band partitioning, multi-scale, and full-band frame modeling.

Ruri-Reranker is a reranking model specifically optimized for Japanese text, based on the sentence-transformers architecture, effectively improving the relevance ranking of search results.

A 7-billion-parameter multilingual translation model based on the Mistral architecture, supporting translation-related tasks in 10 languages

CodeGemma is a series of lightweight open-source code models based on Gemma, focusing on code completion and generation tasks.

CodeGemma is a lightweight open-source code model series based on Gemma, featuring text-to-text and text-to-code decoder models, specializing in code completion and generation tasks.

LLaVA-Gemma-2b is a large multimodal model trained based on the LLaVA-v1.5 framework, utilizing the 2-billion-parameter Gemma-2b-it as the language backbone combined with a CLIP visual encoder.

MusicGen is a text-to-music model that generates high-quality music samples based on text descriptions or audio prompts.

SantaCoder is a 1.1 billion parameter code generation model trained on Python, Java, and JavaScript code, supporting fill-in-the-middle tasks.