bottle.png8.5 mana
Image
GPT Image 2
Meet the new AI Magicx. We’re in closed beta. Public signup is closed.
Beta member sign-inImage · video · voice · music
Image, Video, Voice, Music, Edit, Upscale, Transcribe, Clean voice, Sound. Pick a model, see the mana on the button, keep what you make in your Library. Every paid step is shown before you start.
9 tools · 10 image models
01 · Image
Describe it in plain words. Add reference images to keep a product, a face or a style consistent.
Prices: per image, Good quality, square.
| Model | Mana | Plan |
|---|---|---|
| GPT Image 2OpenAI · Follows long prompts closely and renders text in images well. | ≈ 8.5 | All |
| Gemini 3 Pro ImageGoogle · Google's best image model: layouts, text and consistent characters. | ≈ 21 | Plus |
| FLUX.3 KleinBlack Forest Labs · Very fast, low-cost images with good detail. | ≈ 1.4 | All |
| Seedream 5 FlashByteDance · Crisp, colourful images that follow detailed prompts. | ≈ 4.3 | All |
| FLUX.3Black Forest Labs · Black Forest Labs' best: photoreal detail and clean typography. | ≈ 7.7 | Plus |
| Nano Banana 2.1Google · Quick images and edits with consistent characters and clear text. | ≈ 5.4 | All |
| GPT Image 2.6 FlareOpenAI · Fast drafts at a low price. | ≈ 7 | All |
| GPT Image 2.6 SunburstOpenAI · OpenAI's most detailed images. | ≈ 7 | Plus |
| Ideogram 4.5Ideogram · Clear lettering for posters and logos. | ≈ 9.6 | Plus |
| Ideogram VectorIdeogram · Editable SVG logos and icons. | ≈ 26 | Pro |
02 · Video
Short clips with sound, from a description or from a starting image. Say what moves and how the camera moves.
Prices: per clip, shortest length.
| Model | Mana | Plan |
|---|---|---|
| Gemini Omni 1.2 FlashGoogle · Video with sound, up to 15 seconds and 4K. | ≈ 56 | Pro |
| MiniMax H3 Max TurboMiniMax · Quick, low-cost clips for drafts and social posts. | ≈ 32 | Pro |
| Wan 3.0Alibaba · Natural motion with sound, from words or a starting image. | ≈ 80 | Pro |
| Kling 4.0Kling · Cinematic motion and camera work, up to 1080p. | ≈ 126 | Pro |
| Kling 4.0 FlashKling · Kling motion in clips up to 20 seconds. | ≈ 45 | Pro |
| Seedance 2.5ByteDance · Cinematic video with sound. US-hosted, up to 15 seconds. | ≈ 378 | Pro |
| LTX-2.5 FastLightricks · Video with native sound, up to 4K. | ≈ 86 | Pro |
03 · Voice
Write the script in paragraphs, pick a voice and say how it should sound. Re-record one paragraph without paying for the rest.
Prices: per minute of speech.
| Model | Mana | Plan |
|---|---|---|
| Gemini Flash TTSGoogle · Expressive voices from Google, directed in plain words. | ≈ 4.3 | All |
| Eleven v4ElevenLabs · ElevenLabs' most expressive voices, in 70+ languages. | ≈ 12 | Pro |
| Eleven v4 TurboElevenLabs · Fast, natural ElevenLabs voices at half the price. | ≈ 5.8 | Pro |
| Gemini Flash-Lite TTSGoogle · Low-cost expressive speech with Google's stock voices. | ≈ 2.9 | All |
04 · Music
Describe the mood, instruments and tempo. Instrumental beds or songs with vocals.
Prices: per track, shortest length.
| Model | Mana | Plan |
|---|---|---|
| Lyria 3 ClipGoogle · 30-second clips: jingles, stings and loops. | ≈ 6.4 | Pro |
| Lyria 3.5Google · Full songs or instrumentals from a description, a couple of minutes long. | ≈ 13 | Pro |
| Eleven MusicElevenLabs · Songs and instrumentals made to the exact length you pick. | ≈ 12 | Pro |
| Mureka V9Mureka · Songs with word-timed lyrics. Supply lyrics for a lower price. | ≈ 48 | Pro |
05 · Edit
Pick an image, then say what to change: swap the background, change the colour, add or remove something.
Prices: per edited image.
| Model | Mana | Plan |
|---|---|---|
| GPT Image 2OpenAI · Follows long prompts closely and renders text in images well. | ≈ 11 | All |
| Gemini 3 Pro ImageGoogle · Google's best image model: layouts, text and consistent characters. | ≈ 21 | Plus |
| Nano Banana 2.1Google · Quick images and edits with consistent characters and clear text. | ≈ 5.4 | All |
| GPT Image 2.6 FlareOpenAI · Fast drafts at a low price. | ≈ 9 | All |
| GPT Image 2.6 SunburstOpenAI · OpenAI's most detailed images. | ≈ 9 | Plus |
| Ideogram VectorIdeogram · Editable SVG logos and icons. | ≈ 26 | Pro |
06 · Upscale
Pick an image and how much larger it should be. Detail is sharpened, not stretched.
Prices: per image, 2× larger.
| Model | Mana | Plan |
|---|---|---|
| SeedVR2 UpscaleByteDance · Makes an image 2× or 4× larger with sharp, clean detail. | ≈ 1 | All |
| Topaz PreciseTopaz Labs · Enlarge while preserving the original detail. | ≈ 13 | Plus |
| Topaz RestoreTopaz Labs · Rebuild detail in blurry photos; restored details may differ. | ≈ 13 | Pro |
07 · Transcribe
Add a meeting, interview or voice note. You get a timestamped transcript to copy, edit or download.
Prices: per minute of audio.
| Model | Mana | Plan |
|---|---|---|
| ElevenLabs ScribeElevenLabs · Accurate transcripts with speakers, in 90+ languages, at a low price. | ≈ 1 | Pro |
| GPT-TranscribeOpenAI · Accurate transcripts with language and vocabulary hints. | ≈ 1 | All |
| Gemini 3.5 TranscribeGoogle · Speaker-labelled transcripts, up to 30 minutes. | ≈ 1 | Pro |
08 · Clean voice
Remove background noise while preserving the original.
Prices: per minute of recording.
| Model | Mana | Plan |
|---|---|---|
| ElevenLabs Voice IsolatorElevenLabs · Remove background noise while keeping speech. Original audio is preserved. | ≈ 19 | Pro |
| Quick cleanVEED · A quick speech cleanup; the original is kept. | ≈ 2 | Pro |
09 · Sound
Describe a sound: footsteps on gravel, a door creaking, rain on a tin roof. Loops for ambience.
Prices: per sound, shortest length.
| Model | Mana | Plan |
|---|---|---|
| ElevenLabs Sound EffectsElevenLabs · Footsteps, whooshes, ambience and stingers from a description. | ≈ 1 | Pro |
| Mirelo sound for videoMirelo · Sound effects synced to your video. | ≈ 12 | Pro |
Plan limits
Mana pays for everything. Plans set how much one run can do.
| Limit | Free | Plus | Pro and Max |
|---|---|---|---|
| Images per run | 1 | 4 | 4 |
| Reference images | 2 | 8 | 8 |
| Upscale | 2× | 2× / 4× | 2× / 4× |
| Video | — | — | 720p / 1080p / 1440p / 4k |
| Voice script | 1,000 characters | 5,000 characters | No limit |
| Transcription | 10 minutes | 60 minutes | No limit |
| Music | — | — | Model limits |
More of the kit