DeepSeek-R1-Zero
DeepSeek-R1-Zero is a reasoning large language model from DeepSeek, based on DeepSeek-V3-Base and trained through large-scale reinforcement learning without prior supervised fine-tuning. It is a…
MODEL DIRECTORY
New releases, in one place. Explore models by their publication date.
DeepSeek-R1-Zero is a reasoning large language model from DeepSeek, based on DeepSeek-V3-Base and trained through large-scale reinforcement learning without prior supervised fine-tuning. It is a…
DeepSeek-R1 is DeepSeek's first-generation reasoning model, intended for tasks such as mathematics, programming, and natural language reasoning. The model was officially released on January 20, 2025,…
Tongyi Wanxiang 2.0 Turbo, a text-to-image model provided by Alibaba Cloud Model Studio: it takes text as input and outputs images, suitable for textured portraits, creative design artwork, and other…
Tongyi Wanxiang 2.1 Image-to-Video Plus is a model that converts input images into dynamic videos, with inference services provided through Alibaba Cloud Model Studio. (help.aliyun.com) It supports…
Emoji-Detect is an image detection model used by Alibaba Cloud Model Studio to assist with Emoji sticker video generation. It checks whether the person depicted in an input image meets the…
Emoji is a facial animation video generation model provided by Alibaba Cloud Model Studio. It takes a facial image and a preset animation template as input and generates a corresponding facial…
aitryon-parsing-v1 is an image segmentation support model for OutfitAnyone, the AI virtual try-on service on Alibaba Cloud Model Studio. It can segment model images and clothing images for…
Qwen-Plus qwen-plus-0112 is a text generation snapshot model offered by Alibaba Cloud Model Studio, also known as qwen-plus-2025-01-12. It supports Chinese, English, and other languages, and can be…
A fixed version of Tencent Hunyuan Turbo, corresponding to version pv2.0.6, and a general-purpose large language model. The Turbo series uses a mixture-of-experts (MoE) architecture.…
Wanx2.1-T2I-Plus is the professional edition of the Tongyi Wanxiang 2.1 text-to-image model, with inference services provided by Alibaba Cloud Model Studio. It takes text as input and produces images…
Wanx2.1-T2I-Turbo is a high-speed text-to-image model in Alibaba's Tongyi Wanxiang 2.1 series, with inference services provided through Alibaba Cloud Model Studio. It accepts text prompts and…
Wanx2.1-T2V-Plus is a text-to-video model in Alibaba's Tongyi Wanxiang 2.1 series that generates videos from text prompts. Official descriptions state that it supports complex motion, reproduction of…
Wanx2.1-T2V-Turbo is a high-speed text-to-video model in Alibaba Cloud's Wanxiang 2.1 series that generates videos from text prompts. Official sources state that it supports complex motion, depiction…
Qianfan-Agent-Speed-32K is an agent-specific model developed in-house by Baidu. It is instruction-tuned for enterprise large-model applications and supports a 32K context. Baidu officially positions…
2024-12-26 · DeepSeek A general mixture-of-experts model for conversation, factual questions and coding. The changelog confirms that the deepseek-chat endpoint switched to the original V3 on this date…
2024-12-11 · Google This early experimental Gemini 2.0 release paired the speed of Flash with expanded multimodal and tool capabilities. It was not the following year's stable version. Source
2024-12-05 · OpenAI The full version first appeared in ChatGPT. Access for some API developers followed on December 17; the model identifier does not imply both launches were simultaneous. Source ·…
2024-11-20 · DeepSeek A web preview of a reasoning model for complex problems. It was not the later full R1 release, and this date does not mark the release of R1 weights. The model name stands in for…
2024-11-16 · Moonshot AI Moonshot AI's reasoning model for solving math problems. The model launched on November 16; the Kimi math web experience based on it followed on November 26. The model name…
2024-10-22 · Anthropic This updated Sonnet snapshot improved coding and offered experimental computer use. Its exact snapshot ID distinguishes it from the June version. Source