Compare and access top AI models from OpenAI, Anthropic, Google, BytePlus, Kuaishou and more — all in one place.
BytePlus SeedAudio 1.0 generates up to 120 seconds of audio from text, reference audio, or a reference image for voiceovers, audiobooks, games, and other audio creation workflows.
$0.15 per generated minute, billed by original output duration with one-second precision.
Seedance 2.5 is BytePlus latest-generation video model featuring superior visual fidelity, longer duration support, and advanced scene understanding for professional-grade content creation.
Doubao Seedance 2.5 is the Doubao edition of BytePlus's latest-generation video model, offering superior visual fidelity, longer duration support, and advanced scene understanding for professional-grade content creation.
Seedance 2.5 is BytePlus latest-generation video model featuring superior visual fidelity, longer duration support, and advanced scene understanding for professional-grade content creation.
Seedance 2.0 is BytePlus new-generation video model featuring superior visual fidelity, longer duration support, and advanced scene understanding for professional-grade content creation.
Doubao Seedance 2.0 is the Doubao edition of BytePlus's new-generation video model, delivering strong visual fidelity, longer duration support, and advanced scene understanding for professional-grade content creation.
Seedance 2.0 Normal is the standard BytePlus video generation model, offering balanced visual quality, generation speed, and support for 480p, 720p, 1080p, and 4K output.
Seedance 2.0 is BytePlus new-generation video model featuring superior visual fidelity, longer duration support, and advanced scene understanding for professional-grade content creation.
Pricing varies based on whether the input includes video. 1080p output is not supported.
Doubao Seedance 2.0 Mini is the compact Doubao edition designed for cost-efficient video generation. Pricing varies based on whether the input includes video, and 1080p output is not supported.
Pricing varies based on whether the input includes video. 1080p output is not supported.
Seedance 2.0 Fast is BytePlus fast-response video model built for lower-latency generation, delivering strong visual quality and efficient turnaround for high-throughput creative workflows.
Doubao Seedance 2.0 Fast is the low-latency Doubao edition, delivering strong visual quality and efficient turnaround for high-throughput creative video workflows.
Seedance 2.0 Fast Normal is the standard fast-response BytePlus video model, optimized for lower-latency generation and efficient high-throughput creative workflows.
Seedance 2.0 Fast is BytePlus fast-response video model built for lower-latency generation, delivering strong visual quality and efficient turnaround for high-throughput creative workflows.
Seedance 1.5 is the latest video generation model launched by ByteDance🚀. Based on the functionalities Seedance 1.0 Pro functionalities, Seedance Pro 1.5 can automatically generate matching voices, sound effects, and background music based on text prompts and visual content.
Seedance 1.0 Pro, the large-parameter version of the model suite, features a unique multi-shot storytelling capability and demonstrates outstanding performance across multiple dimensions. It achieves breakthroughs in semantic understanding and instruction-following, enabling the generation of 1080P high-definition videos with smooth motion, rich details, diverse styles, and cinematic visual quality.
Byteplus Seedance 1.0 Professional version, optimized for text-to-video conversion with high-quality video generation capabilities. Supports multiple style transformations, ideal for rapid prototyping and creative exploration. Lightweight design ensures fast response and efficient processing.
BytePlus Seedream 5.0 Pro is the flagship image generation model featuring superior detail rendering, stronger aesthetics, and improved prompt adherence for professional-grade visual creations.
BytePlus Seedream 5.0 Pro is the flagship image generation model featuring superior detail rendering, stronger aesthetics, and improved prompt adherence for professional-grade visual creations.
BytePlus Seedream 5.0 is the latest image generation model featuring enhanced detail rendering, up to 3K/4K resolution output, and improved prompt adherence for stunning visual creations.
BytePlus Seedream 5.0 is the latest image generation model featuring enhanced detail rendering, up to 3K/4K resolution output, and improved prompt adherence for stunning visual creations.
BytePlus Seedream 4.5 is a New High-Aesthetic Image Generation Model with stronger spatial understanding, richer world knowledge, superior aesthetics, higher consistency, and smarter instruction following for precise visual creation.
Byteplus Seedream 4.0 is a SOTA multimodal image generation model, enabling text-to-image creation, image editing and multi-image generation within a single model, supports a wide range of creative scenarios.
Alibaba Cloud Qwen3.8-Max is the latest flagship Qwen text model, delivering top-tier general reasoning, instruction-following and tool use, with built-in web search support.
Alibaba Cloud Qwen3.7-Max is the flagship Qwen text model, delivering strong general reasoning, instruction-following and tool use, with built-in web search support.
Alibaba Cloud Qwen3.7-Plus is the cost-effective Qwen3.7 model, pairing strong text capabilities with upgraded vision-language understanding and full-stack agent intelligence for coding, tool use and multimodal interaction, with a 1M context window.
Alibaba Cloud Qwen3.7-Flash is the lightweight, low-latency Qwen text model optimized for high-throughput and cost-sensitive scenarios while retaining strong general capabilities.
Alibaba Cloud Qwen3.6-Plus is the previous-generation cost-effective Qwen text model with strong text capabilities and tiered pricing based on input length.
Alibaba Cloud Wan 3.0 standard video generation model supporting text-to-video, image-to-video, and reference-based video generation with synchronized audio.
China (Beijing) list prices before the limited-time 30% discount. Input and output video durations are both billable.
Alibaba Cloud Wan 3.0 Prime is the high-speed version of Wan 3.0, delivering the same generation capabilities with significantly faster end-to-end performance.
China (Beijing) list prices. Input and output video durations are both billable.
Google flagship multimodal reasoning model for software engineering, mathematics, science and autonomous agents, with a unified 1M-token context window.
Google native image generation and editing model with 1K/2K/4K output, configurable thinking, search grounding and video-to-image generation.
Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows.
Google's high-speed, efficient Flash model for everyday coding, agentic tool use, and reliable multi-step execution.
Google Flash model balancing speed and multimodal capabilities across general agentic and everyday tasks.
Google's cost-efficient model optimized for high-volume agentic tasks, translation, and simple data processing.
Google's cost-efficient model optimized for high-volume agentic tasks, translation, and simple data processing.
Google Gemini 2.5 Flash model, a fast and efficient language model optimized for speed and cost-effectiveness.
Google Gemini 2.5 Flash model, a fast and efficient language model optimized for speed and cost-effectiveness.
Google Gemini 2.5 Pro model, offering advanced capabilities with enhanced reasoning and understanding.
Google Gemini 2.5 Flash model, a fast and efficient language model optimized for speed and cost-effectiveness.
Google Gemini 3.1 Pro Preview model, the latest preview version with cutting-edge capabilities.
Google Gemini 3.1 Flash-Lite Preview model, optimized for high-throughput lightweight tasks with low latency and cost efficiency.
Google Gemini 3.5 Flash model, optimized for fast multimodal reasoning, search-grounded responses, and cost-efficient high-throughput workloads.
Google's latest Veo 3.1 video generation model. Generate high-quality videos with synchronized speech/sound effects from a text prompt or reference image
Google's latest Veo 3.1 video generation model. Generate videos with synchronized speech/sound effects from a text prompt or reference image faster
Google Gemini Omni Flash Preview supports text-to-video generation, video editing, and multi-turn video editing through the Interactions API.
(Nano Banana 2🍌) Google's lightweight native image generation model built on Gemini 3.1 Flash Lite. Optimized for maximum speed and cost efficiency in high-throughput scenarios while delivering high-quality, photorealistic imagery.
(Nano Banana🍌) Google Gemini 2.5 Flash image model supporting multimodal text and image input/output. Can be used via OpenAI-compatible interface by adding modalities parameter. Lightweight and fast, ideal for real-time application scenarios requiring image understanding and generation.
(Nano Banana Pro🍌) Google Gemini 3 Pro image generation model with advanced multimodal capabilities. Features state-of-the-art image synthesis, enhanced prompt understanding, and superior visual quality. Supports creative image generation, editing, and style transfer with exceptional detail and accuracy.
(Nano Banana 2🍌) is the latest state-of-the-art image model. Dramatically closes the gap between speed and visual fidelity, delivering high-quality, photorealistic imagery.
(Nano Banana Pro🍌) Google's professional native image generation model built on Gemini 3 Pro with a reasoning core, delivering studio-grade 4K visuals, complex layouts and precise text rendering. Text input/output pricing aligned with Gemini 3.1 Pro.
(Nano Banana 2🍌) Google's high-efficiency native image generation model built on Gemini 3.1 Flash. Purpose-built for speed and high-throughput scenarios while delivering high-quality, photorealistic imagery.
OpenAI GPT-6.1 Sol handles complex coding, computer use, and professional tasks, including document analysis and multi-step workflows.
OpenAI GPT-6 Sol is a GPT-6 coding model for demanding software engineering and agentic workflows, with a 1,050,000-token context window and up to 128,000 output tokens.
OpenAI GPT-6 Luna is the smallest and fastest GPT-6 model for cost-sensitive, high-volume general-purpose workloads, with a 1,050,000-token context window and up to 128,000 output tokens.
OpenAI GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation, with a 1,050,000-token context window and up to 128,000 output tokens.
OpenAI GPT Image 2.5 Sunburst is the quality-focused image generation and editing model, designed for precise instruction following and high-fidelity creative and production workflows.
OpenAI GPT Image 2.5 Flare is the faster image generation and editing model, optimized for responsive, high-quality everyday creative workflows.
OpenAI GPT Image 2 is a multimodal image generation and editing model supporting both text and image inputs with high-fidelity image outputs. It delivers strong prompt understanding and instruction following for creative, design, and production scenarios.
xAI's most capable model for coding and knowledge work, with stronger long-horizon reasoning, self-verification, and a 500k token context window.
xAI Grok Imagine image generation model for creating high-quality images from text prompts with 1K and 2K output support.
xAI Grok Imagine quality image model for higher-fidelity image generation with 1K and 2K output tiers.
xAI Grok Imagine video generation model for text-to-video and image-to-video creation with configurable duration and resolution.
xAI's latest frontier model with advanced reasoning and agentic capabilities, featuring tiered pricing for short and long context.
xAI's latest frontier model with advanced reasoning and agentic capabilities, featuring a 500k token context window.
Agentic tool calling, minimal hallucinations, non-reasoning mode with a 1 million token context window.
Kling AI video generation model, optimized for content creation. Supports text-to-video, image-to-video, and multiple generation modes. Features excellent character motion understanding and scene coherence, ideal for short video creation, social media content, and marketing video production.
Kling O1 Video is an advanced video generation model featuring the Omni-Video capability. Supports text-to-video, image-to-video, and video editing with reference images. Offers multiple durations (5-10s), resolutions up to 1080p, and various aspect ratios. Ideal for professional video content creation.
Kling O1 Image is a high-quality AI image generation model. Supports text-to-image with multiple aspect ratios, resolutions (1K/2K), and batch generation up to 9 images. Features excellent prompt understanding and photorealistic output quality for creative and commercial applications.
Kling AI platform providing Omni-Image and Omni-Video generation with multi-subject composition, reference assets, and high-quality media creation.
Vidu Q4 is a reference-to-video model supporting 1-15 reference images, 0-3 reference audio files, synchronized audio-video generation, and 3-16 second videos at resolutions up to 4K.
Off-peak hours are charged at half price. Enabling audio for reference-to-video adds 15 credits ($0.075) per task.
Vidu Q3 Pro is the highest-quality model in the Q3 series, supporting text-to-video, image-to-video, and start-end-to-video with audio-visual synchronization and multi-shot segmentation. Duration 1-16s with resolutions up to 1080P.
Off-peak hours are charged at half price. Enabling audio for image-to-video adds 15 credits ($0.075) per task.
Vidu Q3 Turbo is optimized for faster generation, supporting text-to-video, image-to-video, start-end-to-video, and reference-to-video. Duration 1-16s with resolutions up to 1080P, ideal for quick iterations.
Off-peak hours are charged at half price. Enabling audio for reference-to-video adds 15 credits ($0.075) per task.
Vidu Q3 Mix balances quality and speed for reference-to-video generation in mixed creative scenarios. Duration 3-16s with resolutions up to 1080P.
Off-peak discount is not supported. Enabling audio adds 15 credits ($0.075) per task.
MiniMax-M3.1-Flash is a multimodal coding and agent model with a 1M-token context window, image and video understanding, and always-on thinking with configurable reasoning depth.
MiniMax-H3-Max is a fast video generation model supporting text-to-video and image-to-video workflows with 480P and 768P output.
Only output video duration is billed; input images are currently free.
MiniMax-H3 is a frontier video generation model supporting text-to-video, image-to-video, and video-to-video (with reference audio) workflows, delivering high-fidelity 2K output.
MiniMax-M3 is the latest frontier model from MiniMax, built for agent workflows, coding, reasoning, and complex long-context tasks with up to 1M-token context.
MiniMax-M2.5 is a frontier text model from MiniMax, engineered for agent workflows, coding, reasoning, and complex long-context tasks.
MiniMax Hailuo AI video generation model, renowned for natural fluid motion and precise scene understanding. Supports various video styles from realistic to cartoon, static to dynamic. Built-in intelligent scene analysis automatically optimizes video pacing and transition effects.
GLM-5.3-Flash is the first native multimodal model in the GLM-5 series, combining stronger-than-GLM-5.2 intelligence with efficient coding and agent capabilities, a 1M-token context window, and up to 128K output tokens.
GLM-5.3 is Zhipu's latest flagship model for complex software engineering and agentic tasks, featuring a 1M-token context window and up to 128K output tokens.
GLM-5.2 is Zhipu's flagship model for long-horizon coding and agentic tasks, featuring a usable 1M context window.
GLM-5.1 is a Zhipu GLM-5 series model for coding and agentic tasks, supporting long-context chat completion.
GLM-4.6v is a cost-effective GLM model supporting chat completion.
GLM is Zhipu's flagship model line for complex dialogue and agent tasks. It supports thinking mode, native tool calling, and MCP integration, with up to 128K context for long conversations and multi-step reasoning workflows.
DeepSeek-V4.1-Flash-1017 improves coding and end-to-end project development with faster task completion and shorter outputs. It supports text and image input, thinking and non-thinking modes, a 1M context window, and up to 384K output.
DeepSeek-V4.1-Flash is DeepSeek's fast multimodal model for text and image understanding. It supports non-thinking and thinking modes, a 1M context window, and up to 384K output.
DeepSeek-V4-Flash-0731 is the 0731 snapshot of DeepSeek-V4-Flash, supporting both non-thinking and thinking modes. Context window: 1M. Max output: 384K. Fast and cost-effective for daily tasks and agent workflows.
DeepSeek-V4-Flash supports both non-thinking and thinking modes. Context window: 1M. Max output: 384K. Fast and cost-effective for daily tasks and agent workflows.
DeepSeek-V4-Pro-0813 is the 0813 snapshot of DeepSeek-V4-Pro, the most powerful DeepSeek model with advanced reasoning capabilities. Context window: 1M. Max output: 384K. Ideal for complex reasoning and agent tasks.
DeepSeek-V4-Pro is the most powerful DeepSeek model with advanced reasoning capabilities. Context window: 1M. Max output: 384K. Ideal for complex reasoning and agent tasks.
DeepSeek-V3.2 non-thinking mode for fast chat and standard tool usage. Context window: 128K. Output: default 4K, max 8K.
DeepSeek-V3.2 thinking mode (deepseek-reasoner) for complex reasoning and agent tasks. Context window: 128K. Output: default 32K, max 64K.
Doubao Seed 1.6 Thinking version, optimized for analytical tasks. Features powerful logical reasoning and problem decomposition capabilities, displaying detailed thought processes. Ideal for education, academic research, and business analysis requiring transparent reasoning.
Seed 1.8 general-purpose version by ByteDance Seed team, a balanced language model. Performs well across dialogue, writing, translation, and summarization tasks. Deeply optimized for Chinese scenarios, understanding Chinese context and cultural background, suitable for daily use.
Seed 2.0 Pro (snapshot 260328), the flagship high-capability tier of the BytePlus Seed 2.0 series. Optimized for complex reasoning, writing, translation, and long-context understanding up to 256K tokens.
Seed 2.0 Code Preview (snapshot 260328), a Seed 2.0 tier tuned for coding tasks. Supports context lengths up to 256K tokens.
Seed 2.0 Lite (snapshot 260228), a lightweight Seed 2.0 model supporting text and audio input, with context lengths up to 256K tokens.
Seed 2.0 Lite (snapshot 260428), a lightweight Seed 2.0 model supporting text and audio input, with context lengths up to 256K tokens.
Seed 2.0 Mini (snapshot 260428), the lightest and most cost-effective Seed 2.0 tier, supporting text and audio input with context lengths up to 256K tokens.
Doubao Seed 2.0 Lite (snapshot 260215), a lightweight Seed 2.0 model supporting text and audio input, with context lengths up to 256K tokens.
Claude Haiku 5.5 is built for high-volume, latency-sensitive classification, extraction, routing, and subagent tasks. It supports text and image input, a 1M-token context window, up to 128K output tokens, and adaptive thinking with configurable effort.
USD per million tokens. All rates, including output, are selected by prompt length. Prompts of exactly 100,000 tokens use the lower tier.
Claude Sonnet 5.5 is Anthropic's flagship model for high-performance coding, agentic workflows, and long-context knowledge work. It uses the Anthropic Messages API and supports multimodal input, tool use, and extended thinking.
Claude Opus 5.5 is built for long-running agentic coding and knowledge work. It supports a 1M-token context window, up to 128K output tokens, and always-on adaptive thinking with configurable effort levels.
Claude Fable 5.1 is Anthropic's most capable model for demanding reasoning and long-horizon agentic work, with stronger long-running coding, multistep research, and document workflows. It supports a 1M-token context window, up to 128K output tokens, vision, and always-on adaptive thinking.
Claude Fable 5 is Anthropic's most capable widely-released model, designed for the most demanding reasoning and long-horizon agentic work. Features a 1M-token context window, up to 128K output tokens, always-on adaptive thinking, and vision support. Released June 9, 2026.
Claude Opus 5 is Anthropic's most intelligent model to date, with state-of-the-art performance on complex tasks. It features enhanced reasoning, coding, and agentic capabilities, and supports extended thinking for deeper problem-solving.
Claude Sonnet 5 is Anthropic's latest flagship model, achieving world-leading performance in AI agents, programming, and computer usage. Features enhanced knowledge base and exceptional long-text processing capabilities. Particularly suitable for complex long-term tasks, code review, and technical documentation. Excels in accuracy and attention to detail.
Claude Opus 4.8 is Anthropic's most powerful model, offering exceptional reasoning capabilities, deep knowledge understanding, and nuanced language processing. Designed for the most demanding tasks requiring complex analysis, creative writing, and sophisticated problem-solving.
Claude Opus 4.7 is Anthropic's most powerful model, offering exceptional reasoning capabilities, deep knowledge understanding, and nuanced language processing. Designed for the most demanding tasks requiring complex analysis, creative writing, and sophisticated problem-solving.
Claude Opus 4.6 is Anthropic's advanced reasoning model, designed for complex analysis, coding, and demanding agentic workflows with strong performance and nuanced language understanding.
Claude Sonnet 4.5 is Anthropic's latest flagship model, achieving world-leading performance in AI agents, programming, and computer usage. Features enhanced knowledge base and exceptional long-text processing capabilities. Particularly suitable for complex long-term tasks, code review, and technical documentation. Excels in accuracy and attention to detail.
Claude Sonnet 4.6 is Anthropic's latest flagship model, achieving world-leading performance in AI agents, programming, and computer usage. Features enhanced knowledge base and exceptional long-text processing capabilities. Particularly suitable for complex long-term tasks, code review, and technical documentation. Excels in accuracy and attention to detail.
Claude Haiku 4.5 is a fast, affordable, and highly capable AI model excelling at programming and agentic tasks. Perfectly combines speed with low cost, ideal for real-time applications, large-scale deployments, and scenarios requiring rapid responses. Provides industry-leading cost-effectiveness while maintaining high-quality output.
Claude Opus 4 is a fast, affordable, and highly capable AI model excelling at programming and agentic tasks. Perfectly combines speed with low cost, ideal for real-time applications, large-scale deployments, and scenarios requiring rapid responses. Provides industry-leading cost-effectiveness while maintaining high-quality output.
OpenAI GPT-5.6-sol is the flagship model of the gpt-5.6 family, designed for the most demanding reasoning and agent workloads. Adds the new reasoning.mode=pro, reasoning.effort=max, prompt_cache_options, and programmatic tool calling capabilities.
OpenAI GPT-5.6-terra balances capability and cost for mainstream chat and reasoning workloads across the gpt-5.6 family.
OpenAI GPT-5.6-luna is the low-cost, high-throughput member of the gpt-5.6 family, optimized for lightweight chat and reasoning workloads at scale.
OpenAI GPT-5.5 model for premium long-context and advanced reasoning workloads.
FlashVSR is a fast, high-quality video upscaler that boosts resolution and restores clarity for low-resolution or blurry footage. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.
Kimi K3 is Moonshot AI's flagship model for long-horizon coding and end-to-end knowledge work, with a 1M-token (1,048,576) context window and industry-leading intelligence. It always reasons and supports configuring its reasoning level via the top-level reasoning_effort request field (low / high / max, default max).
Kimi-k2.7-code is Moonshot AI's most capable coding model, delivering reliable instruction-following over long contexts and high task success rates on programming and agentic workloads. Supports text, image, and video input with native chain-of-thought (CoT) output via the reasoning_content field. This endpoint also supports kimi-k2.5, a multimodal model offering both thinking and non-thinking modes.
Tencent Hunyuan's image model for text-to-image generation, reference-image input, and multi-turn editing through the Messages protocol, with up to 4K output.
Hy4 preview is Tencent Hy's 770B-parameter MoE flagship model, activating 49B parameters per token and supporting a 1M-token context window for complex agentic and productivity workloads.
hy3 is Tencent's Hunyuan 3 model, served through an OpenAI-compatible chat completions interface for general reasoning, coding, and agentic tasks.
hy3-preview is Tencent's Hunyuan 3 preview model, served through an OpenAI-compatible chat completions interface. It delivers strong general reasoning and instruction-following across long contexts, and can emit its chain-of-thought via the reasoning_content field alongside the final content.
Sonilo audio model for music and sound effect generation from text or video inputs, plus automatic audio ducking. Billed per second of output audio.