Live
StarHOT · AI Glossary
AI Glossary
Short definitions of emerging models, techniques and products, taken only from stories already published on StarHOT.
- Entries
- 8
- Added this week
- 0
Newly added
A–Z
- FlintFlint is a 40B parameter model developed by AI startup Recursive, designed for code understanding, debugging, feature implementation, and testing.Models9 stories
- GPU UnlockGPU Unlock is an open-source project that unlocks the full AI computing potential of RTX 30 series graphics cards by removing instruction emission rate limits in tensor core operations, potentially increasing performance by up to 60%.Techniques1 story
- H3H3 is a multimodal generation model developed by MiniMax, capable of generating high-resolution video and supporting unified context understanding for text, images, video, and audio.Models9 stories
- Hy-SimulMT-30B-A3BHy-SimulMT-30B-A3B is a real-time simultaneous translation model developed by Tencent Mixuan, supporting bidirectional Chinese-English voice-to-voice translation and real-time voice cloning.Models6 stories
- Hydrogen-1.0Hydrogen-1.0 is an open-source scientific model released by Deep Principle with 753 billion parameters, designed for applications in chemistry, materials science, and life sciences, scoring 38.3% in the FrontierScience-Research evaluation.Models1 story
- LFM3LFM3 is a series of local foundation models developed by Liquid AI, derived from the LFM3-1.7B-Base backbone, supporting text, visual, audio, multimodal, and decision-making tasks, and available as open-weight models on Hugging Face.Models7 stories
- Mistral Large 4Mistral Large 4, also known as Le Chonk, is a multimodal model released by Mistral with approximately 1 trillion total parameters and 49 billion active parameters, offering FP8 and NVFP4 versions, and a context window of 1 million tokens.Models3 stories
- Nanbeige4.3Nanbeige4.3 is a 33B parameter dense large language model released by Nanbeige, supporting a context window of 524,288 tokens and leading in 17 out of 20 benchmark tests.Models1 story
Models
7 entries- FlintModelsFlint is a 40B parameter model developed by AI startup Recursive, designed for code understanding, debugging, feature implementation, and testing.
- H3NEWModelsH3 is a multimodal generation model developed by MiniMax, capable of generating high-resolution video and supporting unified context understanding for text, images, video, and audio.
- Hy-SimulMT-30B-A3BModelsHy-SimulMT-30B-A3B is a real-time simultaneous translation model developed by Tencent Mixuan, supporting bidirectional Chinese-English voice-to-voice translation and real-time voice cloning.
- Hydrogen-1.0NEWModelsHydrogen-1.0 is an open-source scientific model released by Deep Principle with 753 billion parameters, designed for applications in chemistry, materials science, and life sciences, scoring 38.3% in the FrontierScience-Research evaluation.
- LFM3ModelsLFM3 is a series of local foundation models developed by Liquid AI, derived from the LFM3-1.7B-Base backbone, supporting text, visual, audio, multimodal, and decision-making tasks, and available as open-weight models on Hugging Face.
- Mistral Large 4ModelsMistral Large 4, also known as Le Chonk, is a multimodal model released by Mistral with approximately 1 trillion total parameters and 49 billion active parameters, offering FP8 and NVFP4 versions, and a context window of 1 million tokens.
- Nanbeige4.3ModelsNanbeige4.3 is a 33B parameter dense large language model released by Nanbeige, supporting a context window of 524,288 tokens and leading in 17 out of 20 benchmark tests.
