A next-generation cost-efficient flagship workhorse featuring 30%+ faster output, optimized for everyday coding and professional workflows.
A cost-efficient frontier model offering near-Astra intelligence, specializing in complex coding, computer use, and scientific research.
Qwen-Audio-3.0-TTS-Plus is a high-performance large-scale text-to-speech model designed for high-quality speech generation scenarios.
A next-generation high-fidelity speech synthesis model featuring lifelike natural prosody, rich emotional depth, and instant voice cloning.
A next-generation empathetic and creative foundation model specializing in deep resonance, artistic synthesis, and nuanced interaction.
A top-tier cognitive and deep-reasoning model specializing in complex logic analysis, full-stack architecture, and extended thinking.
A next-generation frontier foundation model featuring top-tier scientific insight, long-horizon planning, and autonomous problem-solving capabilities.
A next-generation lightweight, high-efficiency reasoning model specializing in code architecture, mathematical derivation, and rapid response.
Google's most intelligent Flash model, designed for long-running software engineering, autonomous agents, and complex enterprise workflows.
A lightweight, next-generation image model combining ultra-fast rendering with lifelike, detailed visual generation.
A flagship high-fidelity image generation model featuring superior lighting rendering, dynamic colors, and ultra-fine details.
Built for complex reasoning and long-running tasks, excelling at coding, multistep research, and document work.
Doubao Seedream 5.0 is Doubao’s flagship image-generation model, offering stronger knowledge grounding, reference consistency, prompt understanding, and professional visual creation for diverse commercial and creative scenarios across modern production workflows.
Doubao Seed 2.1 Pro is a flagship multimodal reasoning model built for complex analysis, coding, long-context understanding, document processing, and reliable agentic workflows across demanding professional applications and business scenarios.
GPT-6 Astra is OpenAI’s flagship model for demanding end-to-end tasks. It excels at complex reasoning, software engineering, computer use, research, and professional document creation. The model supports text and image input, with a context window of approximately 1.05 million tokens and up to 128,000 output tokens.
Wan 3.0 Video is Alibaba Cloud’s all-in-one multimodal video generation model. It accepts text, images, video, audio, documents, and web links for text-to-video, first-and-last-frame generation, and reference-based video creation. It can natively generate dialogue, background music, and sound effects, with output up to 1080P, 30 FPS, and 30 seconds.
阿里通义 Video
Grok 4.6 is xAI’s flagship model for coding, agentic tasks, and knowledge work. It offers advanced reasoning, code generation, tool use, and long-running task execution. The model supports text and image input, text output, and a 500K-token context window, making it suitable for software development, research, complex analysis, and automated workflows.
Grok Imagine Video 3 is xAI’s video generation model for creating videos from text, images, and reference media, with optional preset voices. It is suitable for advertising, product visualization, creative production, and social media content.
A next-generation bilingual foundation model featuring outstanding long-context analysis, code architecture, and deep-reasoning capabilities.
2.4万亿参数MoE旗舰，编程与办公能力全面跃升，可自主编程十数天交付完整项目
Qwen-Image 3.0 Pro is the flagship model for professional image generation and editing. It offers stronger complex-instruction understanding, realistic textures, fine details, semantic alignment, and text rendering for advertising, branding, infographics, and information-dense layouts.
Kling Video V3 Omni is a professional multimodal video model focused on advanced reference generation and exceptional consistency. It understands text, images, video, and audio while preserving character appearance, voice traits, objects, and visual style across newly generated scenes.
MiniMax Video
Xiaomi’s flagship trillion-parameter MoE model for complex software engineering, tool use, and long-horizon agentic tasks.
A high-performance model for complex reasoning and agentic tasks.
A next-generation hybrid-reasoning multimodal model, combining tunable deep thinking with lightning-fast responsiveness.
Explore Kimi K3 for long-horizon coding, knowledge work, deep reasoning, visual understanding, and a 1-million-token context window.
ByteDance's Seedance 2.5 is a next-generation AI video generation model that supports both text-to-video and image-to-video generation.
Seedance 2.5 is ByteDance’s multimodal video generation model for international markets. It accepts text, image, video, and audio inputs and can generate videos up to 30 seconds long, with reference-based creation, editing, and extension capabilities.
MiniMax旗舰模型，原生深度集成多模态能力，高效处理百万字长文档
Claude Fable 5 是 Anthropic 能力最强的广泛发布模型
Sol（旗舰模型）具备顶级语言理解力，是复杂推理的首选引擎
智谱 GLM-5.2, MIT 开源旗舰全面释放代码潜能，树立自主可控大模型标杆
Seedream 5.0 Lite is ByteDance’s next-generation image generation and editing model for international markets. It offers stronger multimodal understanding, complex instruction following, subject consistency, and detail preservation. The model supports text-to-image, single- and multi-image editing, image blending, and sequential image generation, with output resolutions from 2K to 4K.
Seedream 4.5 is ByteDance’s high-quality image generation and editing model. Compared with Seedream 4.0, it improves editing consistency, preservation of subjects and lighting, portrait quality, and small-text rendering. It supports text-to-image, multi-image blending, image editing, and sequential image creation at 2K or 4K resolution.
Qwen-Image 3.0 is a general-purpose image generation and editing model that balances output quality and response speed. It supports multilingual text rendering and is suitable for posters, web pages, interfaces, product visuals, and everyday creative production.
Kling Video V3 is a professional video generation model supporting videos of up to 15 seconds with native synchronized audio. It delivers improved photorealism, complex motion, multi-shot storytelling, text preservation, and subject consistency for advertising, short films, e-commerce, and cinematic production.
The 5090 Self- deploying general-purpose video generation model balancing visual quality, motion coherence, instruction following, and generation efficiency. Ideal for polished videos, advertising assets, and creative content.
A native omnimodal agent model for long-context reasoning across text, images, video, and audio by XiaoMi.
A fast, cost-efficient model with strong agentic capabilities.
A cost-efficient, ultra-responsive multimodal model featuring millisecond-level inference and robust cross-modal understanding.
Claude Sonnet 5 深度强化逻辑交互，输出结果更精准稳定
Terra（均衡模型）算力与效率精准权衡，广泛适配多元业务流
月之暗面 Kimi,专注Agent开发与代码生成，高效解析超长文本
智谱旗舰, 依托MIT协议开放前沿算法，全力驱动产业智能化升级
The 5090 Self-Deploying speed-optimized video generation model that reduces processing time while maintaining solid visual quality and motion consistency. Ideal for rapid previews, batch production, and frequent creative workflows.
A high-performance model for complex reasoning and agentic tasks.
Claude Opus 5 是 Anthropic 推出的 Opus 级别旗舰模型
Luna（轻量模型）主打低延迟高吞吐，专为高频简易调用场景定制
阿里通义旗舰 Agent 基座, 200 万上下文, 强代码 / 工具调用
字节火山 Seedance 2.0 视频生成, 突破物理动态限制，缔造电影级高清画质视频
Anthropic 旗舰, 突破推理极限兼备，顶尖Agent规划与代码实力
月之暗面 Kimi, 聚焦自动化智能体编排，解锁超长文档深度分析范式
The 5090 self-deploying lightweight and cost-efficient video generation model optimized for rapid creative validation and basic video production. Ideal for short-form videos, draft previews, and high-concurrency workloads.
A fast, cost-efficient model with strong agentic capabilities.
Seedance 2.0 快速版, 低延迟高性价比
OpenAI 新一代旗舰, 强化逻辑推理与工具链，深度重构软件开发工作流
Anthropic Claude Opus 4.7, 专注攻坚复杂编程，推导过程严谨可靠
高速多模态 Gemini, 精妙平衡生成质量与运算成本，助力业务规模落地
通义 3.7 多模态 Agent 模型, 精准识别UI控件并实操，编程调试一体化无缝衔接
字节火山 Seedance 2.0 视频生成, 突破物理动态限制，缔造电影级高清画质视频
深度求索旗舰, 突破代码与数学极限，开放百万字长文本协同框架
阿里 HappyHorse 1.1 文生视频, 实现外语对白口型丝滑同步，有效打破语种传播壁垒
Anthropic Claude Opus 4.6, 持续深化深层推理机制，致力于打造企业级智能中枢
GPT-5.4 轻量版, 大幅优化响应速度，以极高性价比满足日常开发需求
通义前代 Agentic 旗舰, 百万上下文, 深度推动企业工作流智能化
轻量级AI视频生成模型，低成本快速创作高效输出
DeepSeek 高性价比版, 大幅降低Token调用门槛，普惠普及长文本应用
阿里 HappyHorse 1.1 图生视频, 基于首帧解析自动生成对口唇形与背景运镜，交互自然
均衡型 Claude, 实现代码与长文本处理能力的最优平衡，通用首选
海外版 Seedance 2.0 针对性优化本地化特效渲染能力
通义高性价比 flash 档, 低延迟高吞吐, 百万上下文
高速AI视频生成模型，优化响应速度与实时创作体验
阿里 HappyHorse 1.1 参考生视频, 通过多模态特征提取，精准锁定IP角色神态与动作
海外版 Seedance 2.0 快速版, 低延迟高性价比
Seedance 2.0 is an advanced multimodal AI video generation model developed by gobal ByteDance's Seed team.
专业级AI视频生成模型，支持高真实感智能化创作
通义万相 2.7 文生视频, 支持剧本直驱成片，融合多机位运镜与高品质音频混音
轻量多模态模型，支持视听图文输入，专为高并发API与轻量Agent打造
Gemini极速图像模型，极致压缩运营成本，专为大规模商业化部署打造
通义万相 2.7 图生视频, 静态画作秒变流畅影像，保留原作风格并新增声音维度
通义万相 2.7 参考生视频, 注入角色与道具约束，确保跨镜头叙事视觉绝对一致
全能主力模型，兼顾高质量生成与广博知识库，擅长多参考图一致性编辑
OpenAI 图像生成, 输出电影级高清画质，支持精准局部重绘与风格化编辑
通义新一代图像模型, 7B / 原生 2K, 创意生成与版面精修二合一
Gemini 原生图像生成与编辑，覆盖从概念发散到精细打磨的一站式视觉生产
通义图像专业版, 专攻4K出版级排版，精准还原文字细节，杜绝排版错漏
阿里 HappyHorse 1.0 文生视频, 摒弃迭代扩散模式，单次前向传播即刻生成视频
阿里 HappyHorse 1.0 图生视频, 采用极简拓扑网络，将静态海报瞬间转化动态短片
阿里 HappyHorse 1.0 参考生视频, 支持最多五张草图并行约束，严密把控空间透视感
阿里 HappyHorse 1.0 视频编辑, 仅需语音指令干预关键帧，免重渲染实现视频修改