Models

MoonshotAI's flagship open-source model — 1T total / 32B active MoE architecture. Frontier-class reasoning for complex video scripts, agentic editing commands, and long-context analysis. Runs via Ollama (Q3/Q4/Q5 GGUF) or TurboQuant (HuggingFace, 4-bit NF4). Context: 128K tokens.

Kimi Vision-Language model with 3B active parameters. Understands images and text for scene analysis, video frame description, and multimodal editing commands. Two variants: Instruct (general) and Thinking (chain-of-thought reasoning). Runs via TurboQuant with 4-bit NF4 quantization.

Automatic speech recognition with word-level timestamps. Powers transcription, subtitles, and text-based editing. Available from tiny (39M) to large-v3 (1.5B). TurboQuant compression lets Whisper Medium fit where only Base could before.

Powers AI commands, script writing, chapter analysis, clip finding, smart suggestions, and prompt enhancement. Runs via Ollama. Available at 1B (Lite), 3B (Standard), and 8B (Pro) with Q3/Q4/Q5 quantization.

High quality 7B instruction-tuned model. Available at Q4 (Standard tier) and Q5 (Pro tier). Strong at structured JSON generation for editor commands and content analysis.

TurboQuant-validated model family from 0.5B to 14B. The 3B Instruct variant is benchmarked with full compression data (0.9986 cosine similarity at 4-bit). Also includes Coder variants for technical content.

Generate images from text prompts for video overlays, thumbnails, and B-roll. Supports SDXL Turbo (4-step fast generation) and full SDXL (20-step quality).

Next-generation text-to-image with superior prompt adherence and image quality. Alternative to SDXL for higher quality results.

Local voice synthesis with voice cloning from a 6-second sample. Supports 17 languages. TurboQuant at 3-bit shrinks it from 1.8 GB to 0.6 GB while preserving voice quality.

Purpose-built for 22 Indian regional languages. Saaras v3 for transcription, Bulbul v3 for text-to-speech with 37+ natural voices, plus translation and transliteration. Hindi, Tamil, Telugu, Kannada, Bengali, Malayalam, and more.

Ultra-fast cloud TTS and STT. Lightning TTS with 80+ voices across 15 languages at ~100ms latency. Pulse STT covers 39 languages with speaker diarization and emotion detection.

Neural speaker diarization that identifies who is talking when. Powers multi-speaker detection and automatic speaker-boundary cuts. Falls back to FFmpeg silence-based detection when unavailable.

Real-time face detection for auto-reframe. Tracks faces across video frames to generate 9:16 crops for TikTok, Reels, and Shorts.

Detects emotional peaks in audio for finding the most impactful moments. Used for clip scoring and highlight detection. Falls back to FFmpeg energy-based analysis locally.

KV cache compression using Google Research's PolarQuant + QJL. Achieves 6x memory reduction at 3-bit with 0.9953 cosine similarity. Makes 7B models run on 8 GB RAM. One-click setup in Settings.

Remove backgrounds from images to create transparent overlays. Useful for speaker cutouts, product shots, and compositing.

Spectral gating noise reduction for cleaning up audio recordings. Removes background noise, hiss, and hum with adjustable strength.