MediaLab
Video, audio and editing in one section
Hailuo-02 (MiniMax) and Kling v2.1 (Kuaishou) for video, Nano Banana for editing images without a designer's background, VibeVoice from Microsoft for voiceover.
Models by section
Image editing
Nano Banana edits a picture from a plain instruction: “remove the man on the right”, “make the background night-time”. Recraft removes backgrounds.
Video from text or a still
Hailuo-02 for short clips. Kling v2.1 for longer shots with motion control and a strong cinematic look.
Voiceover
VibeVoice (Microsoft) — text-to-speech with adjustable timbre and intonation. Duration is billed separately.
Reasoning tool
Phi-4 Reasoning Plus — a specialized model offered as a standalone tool, not for the main chat.
What MediaLab does not do — common questions
Sora
No. Use Hailuo-02 or Kling v2.1 instead.
Suno / Udio (music generation)
No. VibeVoice text-to-speech only.
Whisper (speech recognition)
No. Speech-to-text is on the roadmap.
Midjourney / Flux / Stable Diffusion
No. Images from scratch are made in the regular chat with GPT-5 Image or Nano Banana Pro.
Ready to generate your first video?
You pay per result: a fixed number of neurons per clip or per second of voiceover. If it fails for technical reasons, you are refunded automatically.