Image, video and audio
Image, video, voice, speech-recognition and music models and products, newest first. LMArena rates images and video only; no voice or music leaderboard publishes its data under a licence we may use. Parameters: the size the maker states, or the total of the weights in the maker's own Hugging Face repository (point at the number for its source). Closed models do not publish their size.
Refreshed Nov 2, 2026, 5:25 AM PT We refresh the data every morning.
LMArena published Oct 26, 2026; Hugging Face read Nov 2, 2026; Makers' own pages checked Oct 26, 2026.
/
From our news
| Model or product | Maker | Kind | Latest version | Released | Weights | Parameters | LMArena | Downloads, 30 d | Our Shorts |
|---|---|---|---|---|---|---|---|---|---|
| Voxtral | Mistral | Speech recognition | Voxtral-Mini-4B-Realtime-2602; Voxtral 3 API preview 2026-10-20 | Oct 20, 2026 | open | – | – | – | – |
| Klingpreview | Kuaishou | Video | Kling 4.0 (API early access) | Oct 15, 2026 | closed | 1220 video from text 1354 video from an image 1255 video editing | – | – | |
| FLUX 3 Image Klein | Black Forest Labs | Images | flux-3-image-klein | Oct 13, 2026 | open | 5.6B | 1070 image from text 1224 image editing | 155,600 | Oct 15 |
| Vidupreview | Shengshu | Video | Vidu Q4 Preview | Oct 7, 2026 | closed | 1363 video from an image | – | – | |
| Nano Banana | Images | gemini-nano-banana-2.1 (GA) | Oct 6, 2026 | closed | 1325 image from text 1427 image editing | – | – | ||
| FLUX 3 Image (API) | Black Forest Labs | Images | FLUX 3 Image | Oct 1, 2026 | closed | – | – | Oct 15 | |
| Fish Audio | Fish Audio | Voices | Drama 3 (2026-10-01); S2 Pro (2026-03-09) | Oct 1, 2026 | open | 35B total3B active | – | 35,078 | Oct 13 |
| ElevenLabs TTS | ElevenLabs | Voices | eleven_v4 / v4_turbo | Sep 28, 2026 | closed | – | – | – | |
| Gemini TTS | Voices | gemini-3.8-flash-tts (GA) | Sep 22, 2026 | closed | – | – | – | ||
| Qwen-Image | Alibaba Qwen | Images | Qwen-Image-2.1 | Sep 14, 2026 | open | 7B | 1254 image from text 1362 image editing | 223,194 | – |
| Eleven Music | ElevenLabs | Music and sound | music_v2_5 | Sep 14, 2026 | closed | – | – | – | |
| OpenAI voice | OpenAI | Voices | gpt-live-1 (GA 2026-09-10); gpt-realtime-2.1 (2026-07-06); gpt-4o-mini-tts shuts 2027-01-06 | Sep 10, 2026 | closed | – | – | – | |
| Kandinsky | Kandinsky Lab | Video | Kandinsky 6.0 | Sep 9, 2026 | open | 29Bsizes 3B to 29B | 1175 video from text | 547 | – |
| Suno | Suno | Music and sound | v6 (v6-wild, v6-mini) | Sep 9, 2026 | closed | – | – | – | |
| YuE2 | M-A-P | Music and sound | YuE2-3B | Sep 9, 2026 | open | – | – | 23,691 | – |
| GPT Image | OpenAI | Images | gpt-image-2.5-flare / -sunburst | Sep 8, 2026 | closed | 1427 image from text 1517 image editing | – | – | |
| Cartesia Sonic | Cartesia | Voices | sonic-3.6 | Aug 27, 2026 | closed | – | – | – | |
| MAI-Image | Microsoft | Images | MAI-Image-2.6 | Aug 10, 2026 | closed | 1329 image from text 1471 image editing | – | – | |
| IndexTTS | Bilibili IndexTeam | Voices | IndexTTS-2.5 | Aug 10, 2026 | open | 800M | – | 23,951 | – |
| Grok Imagine (image) | xAI | Images | grok-imagine-image-2.0 | Aug 7, 2026 | closed | 1317 image from text 1439 image editing | – | – | |
| MiniMax Music | MiniMax | Music and sound | Music-3.0 | Aug 7, 2026 | open + API | – | – | 5,976 | – |
| Seedance | ByteDance | Video | Seedance 2.5 | Jul 31, 2026 | closed | 1479 video from text 1477 video from an image 1410 video editing | – | – | |
| Lyria | Music and sound | lyria-3.5 (Flow Music 2026-07-29; Gemini API GA 2026-09-03) | Jul 29, 2026 | closed | – | – | – | ||
| MiniMax H3 (Hailuo 3) | MiniMax | Video | MiniMax-H3 | Jul 28, 2026 | open | 33B | 1460 video from text 1495 video from an image 1392 video editing | 6,140,874 | – |
| Midjourney | Midjourney | Images | V8.2 (default) | Jul 24, 2026 | closed | – | – | – | |
| LTX | Lightricks | Video | LTX-2.5 | Jul 23, 2026 | open | 22B | 1154 video from text 1159 video from an image | 2,467,613 | – |
| Reve | Reve | Images | Reve 2.1 | Jul 9, 2026 | closed | 1300 image from text 1373 image editing | – | – | |
| Seedream | ByteDance | Images | Seedream 5.0 Pro | Jul 8, 2026 | closed | 1250 image from text 1390 image editing | – | – | |
| Muse Image | Meta | Images | muse-image-1.0 | Jul 7, 2026 | closed | 1269 image from text 1400 image editing | – | – | |
| Grok Imagine (video) | xAI | Video | grok-imagine-video-1.5 | Jun 16, 2026 | closed | 1492 video from text 1456 video from an image 1258 video editing | – | – | |
| Luma Ray | Luma | Video | Ray3.2 | Jun 9, 2026 | closed | 1206 video from text 1226 video from an image | – | – | |
| Higgs TTS | Boson AI | Voices | higgs-tts-3-4b | Jun 4, 2026 | open | 4B | – | 123,465 | – |
| Magenta RealTime | Music and sound | magenta-realtime-2 | May 28, 2026 | open | 2.4Bsizes 230M to 2.4B | – | 29,831 | – | |
| Runway | Runway | Video | Gen-4.5 (2025-12-11); Aleph 2.0 (2026-05-21) | May 21, 2026 | closed | 1225 video from text 1052 video from an image 1182 video editing | – | – | |
| Stable Audio 3 | Stability AI | Music and sound | Stable Audio 3.0 | May 20, 2026 | open + API | – | – | 83,127 | – |
| NVIDIA Parakeet / Canary / Nemotron ASR | NVIDIA | Speech recognition | parakeet-tdt-0.6b-v3; canary-1b-v2; nemotron-3.5-asr-streaming-0.6b (2026-05-15) | May 15, 2026 | open | 600M | – | 585,304 | – |
| Recraft | Recraft | Images | V4.1 | May 14, 2026 | closed | 1168 image from text | – | – | |
| HiDream-O1-Image | HiDream | Images | HiDream-O1-Image | May 8, 2026 | open | 8B | 1115 image from text | 4,725 | – |
| VoxCPM | OpenBMB | Voices | VoxCPM2 | Apr 3, 2026 | open | 2B | – | 969,017 | – |
| ACE-Step | ACE Studio / StepFun | Music and sound | ACE-Step 1.5 (2026-01-23), 1.5 XL (2026-04-02) | Apr 2, 2026 | open | 4Bsizes 2B to 4B | – | 93,154 | – |
| PixVerse | PixVerse | Video | V6 | Mar 30, 2026 | closed | 1240 video from text 1299 video from an image | – | – | |
| OmniVoice | k2-fsa | Voices | OmniVoice | Mar 30, 2026 | open | – | – | 2,928,673 | – |
| Cohere Transcribe | Cohere | Speech recognition | cohere-transcribe-03-2026 | Mar 24, 2026 | open | 2B | – | 93,722 | – |
| Qwen3-ASR | Alibaba Qwen | Speech recognition | Qwen3-ASR-1.7B / 0.6B | Jan 28, 2026 | open | 1.7Bsizes 600M to 1.7B | – | 1,579,322 | – |
| MiniMax Speech | MiniMax | Voices | speech-2.8-hd/-turbo | Jan 23, 2026 | closed | – | – | – | |
| Qwen3-TTS | Alibaba Qwen | Voices | Qwen3-TTS-12Hz 1.7B/0.6B (Base, CustomVoice, VoiceDesign) | Jan 21, 2026 | open | 1.7Bsizes 600M to 1.7B | – | 1,719,207 | – |
| Qwen-Image-Edit | Alibaba Qwen | Images | Qwen-Image-Edit-2511 | Dec 17, 2025 | open | – | 1241 image editing | 209,787 | – |
| CosyVoice | Alibaba FunAudioLLM | Voices | Fun-CosyVoice3-0.5B-2512 | Dec 11, 2025 | open | 500Msizes 500M to 1.5B | – | 262,996 | – |
| HunyuanVideo | Tencent | Video | HunyuanVideo-1.5 | Nov 18, 2025 | open | 8.3B | 1169 video from text 1198 video from an image | 1,368 | – |
| Veo | Video | Veo 3.1 (GA 2025-11-17; 3.1 Lite 2026-03-31) | Nov 17, 2025 | closed | 1364 video from text 1398 video from an image | – | – | ||
| Hume Octaveclosing | Hume | Voices | Octave 2; TTS and EVI APIs end 2026-11-13 | Oct 3, 2025 | closed | – | – | – | |
| Soradiscontinued | OpenAI | Video | Sora 2 - DISCONTINUED (app 2026-04-26, API 2026-09-24) | Sep 30, 2025 | closed | 1368 video from text | – | – | |
| HunyuanImage | Tencent | Images | HunyuanImage-3.0 (+Instruct) | Sep 25, 2025 | open | 80B total13B active | 1151 image from text 1302 image editing | 1,257 | – |
| Wan 2.2 (open) | Alibaba Wan-AI | Video | Wan2.2 (+Animate-2 2026-07-14) | Jul 24, 2025 | open | 27B total14B active | 1337 video from text 1428 video from an image | 8,363 | – |
| Chatterbox | Resemble AI | Voices | chatterbox (flash 2026-05-28) | Apr 24, 2025 | open | 500M | – | 2,203,606 | – |
| Kokoro | hexgrad | Voices | Kokoro-82M v1.0 | Dec 26, 2024 | open | 82M | – | 11,366,340 | – |
| Whisper | OpenAI | Speech recognition | large-v3 (2023-11); large-v3-turbo (2024-10) | Oct 1, 2024 | open | 809M | – | 5,664,908 | – |
| FLUX.2 dev / klein | Black Forest Labs | Images | FLUX.2-dev, FLUX.2-klein-4B/9B | 2025-11-22 / 2026-01-14 | open | 32B | 1161 image from text 1261 image editing | 330,147 | Oct 15 |
| Ideogram | Ideogram | Images | Ideogram 4.5 | date unverified | open + API | – | 1200 image from text 1347 image editing | – | – |
| Firefly Image | Adobe | Images | Image Model 5 | GA by 2026-03-19 | closed | – | – | – | |
| Krea | Krea | Images | Krea 3 | 2026-10-15/16 | open + API | – | 1128 image from text 1411 image editing | – | – |
| Z-Image | Tongyi-MAI (Alibaba) | Images | Z-Image-Turbo / Z-Image | 2025-11-25 / 2026-01-23 | open | 6B | 1082 image from text | 744,362 | – |
| Gemini Omni | Video | gemini-omni-1.1-flash | GA 2026-08-27 | closed | 1516 video from text 1488 video from an image 1367 video editing | – | – | ||
| Wan 3.0 | Alibaba | Video | wan3.0-video | GA 2026-08-24 | closed | 1476 video from text 1480 video from an image 1414 video editing | – | – | |
| FLUX 3 Video | Black Forest Labs | Video | FLUX 3 Video | API preview 2026-08-04 | closed | 1493 video from text 1449 video from an image | – | – | |
| Muse Videopreview | Meta | Video | preview, not released | announced 2026-07-07 | closed | 1456 video from text | – | – | |
| MAI-Voicepreview | Microsoft | Voices | MAI-Voice-2.1 (preview) | catalog 2026-09-19 | closed | – | – | – | |
| Qwen TTS (API) | Alibaba | Voices | qwen-audio-3.1-tts-plus/-flash | date unverified | closed | – | – | – | |
| Grok voice | xAI | Voices | TTS GA (Mar 2026); grok-voice-3.0 (Oct 2026) | 2026-10 | closed | – | – | Oct 14 | |
| VibeVoice | Microsoft | Voices | VibeVoice-1.5B; Realtime-0.5B | 2025-08-25 / 2025-12-04 | open | – | – | 203,968 | – |
| Closed ASR APIs | OpenAI / Google / ElevenLabs / Deepgram / AssemblyAI / Speechmatics | Speech recognition | gpt-transcribe (2026-07-28); gemini-3.5-transcribe (2026-08-26); Scribe v2 (2026-01-09); Nova-3 + Flux (update 2026-10-23); Universal-3.5 Pro (Sep 2026); Melia 1 (2026-10-13) | 2026 | closed | – | – | – | |
| Udio | Udio | Music and sound | no new model since 2025; licensed app not launched | - | closed | – | – | – |
Sources: our reference list, each checked on the maker's page · LMArena dataset CC BY 4.0 (image and video ratings) · Hugging Face (30-day downloads of the official repository).