CNAPS Studio is a platform that hosts AI models for image, video, and audio tasks and lets users run them in workflows. This monthly ranking looks at which models were run most often on the platform.
In October 2026, MiniMax-H3 Ref2VA ranked first by execution count, followed by Text to Sound Effect (MOSS-SoundEffect v2) and Qwen3 Voice Design. Five models entered the top 10 compared with September.
MiniMax-H3 Ref2VA moves into first place
MiniMax-H3 Ref2VA rose three positions from fourth in September to first in October. It was the highest-ranked video-generation model in this month’s list.
Video upscaling also had two new entries: FlashVSR v1.1 at No. 5 and SeedVR2 7B at No. 7. Their inclusion puts both video creation and video enhancement in the top 10.
Sound effects, voice, and music take three of the top six spots
Text to Sound Effect (MOSS-SoundEffect v2) entered at No. 2, and Qwen3 Voice Design entered at No. 3. Text to Music (ACE-Step 1.5) moved from seventh to sixth.
Those three entries cover different audio tasks: making sound effects, producing speech, and generating music. All three were among the six most-run models on CNAPS Studio in October.
FireRed-1.1 and Qwen-Image-Edit-2511 keep image editing in the mix
FireRed-1.1 finished fourth after leading September’s ranking. Qwen-Image-Edit-2511 climbed two places to eighth. In image generation, Z-Image-Turbo moved from eighth to tenth.
These are relative positions. A model can drop in rank even if its execution count increases, so the arrows should not be read as changes in absolute usage.
Gemma 4 31B entered at No. 9, bringing a vision-language model into October’s top 10.
Running these models on CNAPS Studio
CNAPS Studio hosts the models listed above and provides a place to run them as part of AI workflows. You can start from a prepared example on Home or work with models in Studio to build a workflow for your task.
The ranking can help you discover a model to try. For an actual project, start with the job: editing an image, generating a clip, upscaling footage, creating a voice, or adding sound. Then review the output against what you need.
Explore CNAPS Studio’s ready-to-run examples →
About this ranking
What does “most-used” mean here?
It means the highest number of model executions within CNAPS Studio during October 2026. The ranking describes activity on this platform, not usage across all AI services.
Is this a ranking of the best AI models?
No. Execution count measures usage. It does not measure output quality, speed, or suitability for a particular task.
What does “New” mean?
The model was not in September’s top 10. Five models meet that definition this month: MOSS-SoundEffect v2, Qwen3 Voice Design, FlashVSR v1.1, SeedVR2 7B, and Gemma 4 31B.
