HOMO-1
The cheapest, fastest, most lifelike real-time AI avatar ever shipped.
$0.005 / min · 300 ms · 99% lip-sync.
// 13 personas, generated from a single Imagen-4-Ultra render each
Click any face to swap the avatar above.
Same model, different identity. Bring your own — HOMO-1 doesn't need re-training.
Three numbers that make every other avatar API obsolete.
Every micro-expression.
Driven by your scroll.
Real Veo-generated footage of HOMO-1 — every blink, breath, and lip movement scrubs to your scroll position. In production, the same fidelity runs at 30fps, in real time, for half a cent per minute.
Lip-sync so good, your brain stops noticing.
We aligned to phonemes, not just amplitude. The mouth shapes the word before the sound arrives — exactly like a human.
Live lip-sync · Mediterranean
"real time, real human, real cheap"
Architecture
Neural face mesh — 18ms render budget per frame
Phoneme-first
We predict mouth shapes from phonemes, not waveforms. Same accuracy across English, Mandarin, Greek, Arabic.
Sub-frame timing
Mouth opens 1 frame before the consonant fires. The brain reads it as natural anticipation.
Micro-expressions
Eye blinks, brow lifts, and breath driven by the same prosody model — not a separate animation track.
Drop in one photo.
Talk to it in 8 seconds.
No fine-tuning. No model training. No waiting overnight for a rendered persona. Upload a single still — a selfie, a stock photo, a brand mascot — and HOMO-1 turns it into a fully expressive, real-time speaking avatar instantly.
- Single image input — JPG, PNG, even a cropped passport photo
- Identity-locked: the face stays consistent across every frame
- Ready to talk in 8 seconds, not 8 hours
- No personal voice or video data ever leaves your account
Real-time TTS that clones the speaker's voice.
Pair HOMO-1 with our built-in zero-shot voice cloning. Three seconds of audio is enough to capture intonation, accent, and timbre — then every word the avatar says comes out in that voice, in real time, with sub-300ms turnaround.
- Zero-shot — no training, just a 3-second reference clip
- Synchronised with the face: phonemes drive both lips and waveform together
- 70+ languages with native accent preservation
- On-device opt-in for sensitive use cases (healthcare, banking)
Studio-grade output.
Zero retouching needed.
Macro-photoreal
Iris fibres, eyelashes, micro-pores — every frame holds up to a 4K crop.
Broadcast lighting
Cinematic two-tone rim, anamorphic feel, 16-second seamless loop. Drop straight into a campaign.
Eleven different visual languages.
All HOMO-1.
From liquid chrome to mirror shards to data tendrils — the same model, eleven different art-direction prompts. Drop straight into a campaign, a hero, an ad. No comp shots, no re-renders.
Hover any tile to play.
Particle assembly
Millions of cyan particles converge into a face.
Liquid chrome
Mercury rises and forms the head — surface refraction live-rendered.
Crystal turntable
Translucent glass head slowly rotates, light refracting into rainbows.
Multi-identity morph
One model, five identities, eight seconds.
Glitch reveal
Datamosh resolves into perfect photoreal stability.
3D scan reveal
A laser line passes — wireframe becomes flesh as it goes.
Data tendrils
Glowing ribbons of intent flow around the head.
Beauty turn
Slow 180° in editorial high-key lighting.
Lens flare bloom
Brilliant white bloom resolves into a quiet half-smile.
Mirror shards
Suspended glass fragments orbit the head, reflecting it from every angle.
Water-drop form
A single droplet splashes upward and forms a face mid-air.
Origami unfold
A folded paper square unfolds into a sculpted face.
Ink bloom
A drop of ink in white water blooms into a silhouette.
DNA helix
A double helix tightens upward into a fully formed head.
Statue awakening
A marble bust softens, gains color, opens its eyes.
Smoke condensation
Wisps of cyan smoke condense into solid form, then dissolve.
X-ray → skin
A skull X-ray gradually grows muscle and skin, anatomical reveal.
Voxel buildup
Cubes assemble into a low-poly head, then smooth into photoreal.
Constellation
Star points connect into a face, then flood with light.
Fractal zoom
An infinite Mandelbrot zoom resolves into a HOMO-1 portrait.
Frost crystal
Frost grows into facial contours, then thaws to warm skin.
Brushstroke painting
An invisible brush paints a face into existence on white canvas.
Slow dolly portrait
An 8-second cinematic dolly-in, museum-portrait energy.
Eye-contact macro
Both eyes lock onto camera, blink slowly twice. 4K intimacy.
Whisper
A side mouth whispers a secret — perfect lip-sync at ASMR scale.
Petal storm
A cyclone of rose-gold petals freezes into facial form.
Light prism
A white beam refracts into rainbow that assembles a head.
Neural firing
Neural connections pulse inside a translucent head.
Magnetic sand
Iron sand grains rise and align into a sculpted head.
Multi-angle clone
Three angles of the same face cross-fade in choreographed unison.
Wireframe → skin
A blue mesh fills with skin pixel-by-pixel, ending photoreal.
A face built
from a million voices.
HOMO-1 was trained on the exact moments humans connect: the half-smile, the flicker of recognition, the breath before a sentence.
The price disruption is the product.
Real-time avatars used to be a luxury reserved for $50k pilots. We took an axe to the cost stack.
Competitor list-prices as of Q1 2026. We promise nothing about theirs, everything about ours.
Boring specs.
Magical product.
A custom render pipeline tuned for one job: keep latency under the human conversational threshold (~400ms) at a price that makes shipping it a no-brainer.
Cheap enough to put a human face on everything.
AI Receptionists
Answer calls 24/7 with a face customers actually want to look at. Pennies per call instead of dollars.
Global Sales Demos
A live, lip-synced product demo in any language, on any landing page, for less than a fifth of a cent per minute.
In-game NPCs
300ms means real conversation, not pre-recorded dialogue trees. Ship characters that listen and respond.
Localized Onboarding
70+ languages with native phoneme accuracy. One avatar, every market, no re-shoots.
Talk to it.
You'll never go back.
5 free minutes, no credit card. After that — half a cent a minute, billed by the second. Cancel any time, although you won't.
Humanlike · HOMO-1 · v1.0 · April 2026