Model Series
Explore model series and compare their image, video, and audio models.

CosyVoice is an open-source multilingual TTS model that generates natural, human-parity speech with zero-shot voice cloning in real time.
4 models
FunASR is an open-source, industrial-grade speech recognition toolkit that turns audio into accurate text, with built-in VAD, punctuation, and speaker verification.
1 models
Gemini Transcribe — Google's dedicated ASR model that turns speech into precise, speaker-labeled text in 85+ languages, in real time.
1 models
Gemini Omni — Google's natively multimodal "any-in, any-out" model that turns text, images, audio, and video into editable, physics-aware video, with image and audio outputs on the way.
9 models
Gemini TTS — Google's native speech model that turns text (and a style prompt) into expressive, multi-speaker audio, in Flash and Pro flavors.
3 models
GPT Image — OpenAI's native image generation model that turns text prompts into precise, high-fidelity images with strong prompt-following and accurate text rendering.
6 models
Grok Imagine is xAI's cross-modal generative model that creates and edits images and videos from text, image, or video input, with native synchronized audio.
14 models
Grok Voice is xAI's real-time conversational speech model that lets Grok listen and talk back naturally, with expressive, low-latency, multi-persona voice interaction.
2 models
Hailuo is MiniMax's AI video generation model that turns text or image prompts into smooth, cinematic, physically believable clips with strong motion and character consistency.
6 models
HappyHorse is Alibaba's open-source AI video generation model that produces short clips with synchronized audio in a single pass, topping text-to-video and image-to-video leaderboards.
7 models
Kling is Kuaishou's AI video generation model that turns text or image prompts into high-quality, realistic, cinematic videos with strong motion and character consistency.
23 models
MAI-Transcribe is Microsoft's multilingual speech-to-text model that delivers fast, highly accurate transcription even in noisy, overlapping, or low-quality audio, and ranks first on the FLEURS benchmark.
1 models
MAI-Image is Microsoft's in-house text-to-image model that turns text prompts into high-quality, photorealistic images and currently ranks among the top image generation models.
4 models
MiniMax's video generation models are a family of multimodal video systems that turn text, images, and audio into high-quality, cinematic clips with native synchronized sound, strong motion, and up to 2K resolution.
8 models
MiniMax Speech is MiniMax's text-to-speech model family that generates natural, expressive, multilingual speech with zero-shot voice cloning and controllable emotion and pace.
3 models
Nano Banana is Google's Gemini image generation and editing model that turns text prompts into images and performs precise, conversational edits while keeping characters and scenes consistent.
10 models
PixVerse is AIsphere's AI video generation model that turns text or image prompts into stylized, high-quality videos with strong motion and effects, from realistic clips to anime styles.
13 models
Qwen Audio is Alibaba's multimodal audio-language model that understands speech, sound, and music and can generate text or spoken audio responses across multiple languages.
2 models
Qwen Image is Alibaba's text-to-image model that turns prompts into high-quality images with strong prompt following and accurate, well-rendered text, especially bilingual Chinese and English.
4 models
Seedance is ByteDance's video generation model that turns text or image prompts into smooth, cinematic, multi-shot videos with strong motion, realism, and character consistency.
13 models
Seedream is ByteDance's text-to-image model that turns prompts into high-quality, photorealistic images with strong prompt following, accurate text rendering, and solid bilingual Chinese and English support.
7 models
Veo is Google DeepMind's video generation model that turns text or image prompts into high-quality, cinematic videos with native synchronized audio and strong physical realism.
6 models
Vidu is Shengshu Technology's video generation model that turns text or image prompts into high-quality, realistic videos with strong reference-to-video consistency and coherent motion.
24 models
Wan is Alibaba's open-source video generation model family that turns text or image prompts into high-quality, cinematic videos with strong motion, physics, and native audio.
11 models
Whisper is OpenAI's open-source speech recognition model that transcribes and translates multilingual audio into accurate text with strong robustness to noise and accents.
1 models
Z-Image is Alibaba Tongyi's efficient open-source text-to-image model that turns prompts into high-quality photorealistic images with accurate bilingual Chinese and English text rendering at fast speed.
1 models