# Spacelike AI

Spacelike AI GmbH is a Vienna-based studio that builds production-grade,
performance-optimised AI-model inference for dedicated AI accelerators.

## Company

- Legal name: Spacelike AI GmbH
- Founded by: Friedrich Schöller, Malte Fiala
- Address: Susi-Nicoletti-Weg 9/22, 1100 Vienna, Austria
- Contact: mail@spacelike.ai
- Firmenbuchnummer: FN 610679 h
- VAT ID: ATU79793378

## What we do

We implement and performance-optimise open-weight AI models for
purpose-built AI hardware. Current implementations include:

- FLUX.1 [schnell] — Black Forest Labs, 12 B-param rectified-flow
  transformer, distilled to 1-4 sampling steps. Apache 2.0.
- FLUX.2 [dev] — Black Forest Labs. FLUX.2 Non-Commercial License.
- Stable Diffusion 3.5 Medium and Large — Stability AI. Community
  License.
- Qwen-Image — Alibaba Qwen team, 20 B-param multimodal diffusion
  transformer with accurate in-image text rendering. Apache 2.0.
- Motif — MIT.
- Mochi 1 — Genmo AI, 10 B-param AsymmDiT video diffusion. Apache 2.0.
- Wan 2.2 — Alibaba. Apache 2.0.

See https://spacelike.ai/ for live samples and per-model specs at
https://spacelike.ai/models/<id>/.

## Per-model specifications (Markdown)

Clean Markdown for each implemented model — front matter, full spec, and
sample-output reference — generated for LLM / agent ingestion:

- [FLUX.1 \[schnell\]](https://spacelike.ai/models/flux-1-schnell.md): Distilled rectified-flow transformer · 12 B parameters · Black Forest Labs
- [FLUX.2 \[dev\]](https://spacelike.ai/models/flux-2-dev.md): Latent flow matching + VLM · 32 B parameters · Black Forest Labs
- [Mochi 1](https://spacelike.ai/models/mochi-1.md): AsymmDiT video diffusion · 10 B parameters · Genmo AI
- [Motif](https://spacelike.ai/models/motif.md): MMDiT + flow matching · 6 B parameters · Motif Technologies
- [Qwen-Image](https://spacelike.ai/models/qwen-image.md): Multimodal Diffusion Transformer · 20 B parameters · Alibaba Qwen Team
- [SD 3.5 Large](https://spacelike.ai/models/sd-3-5-large.md): Multimodal Diffusion Transformer · 8.1 B parameters · Stability AI
- [SD 3.5 Medium](https://spacelike.ai/models/sd-3-5-medium.md): Multimodal Diffusion Transformer · 2.5 B parameters · Stability AI
- [Wan 2.2](https://spacelike.ai/models/wan-2-2.md): MoE video diffusion · 27 B total / 14 B active · Alibaba Tongyi Lab

## Engagement model

We work directly with hardware vendors (silicon and accelerator-card
makers) on implementing and optimising open-weight image and video
models on their stacks.

## Image licensing

Sample images on the site are licensed under Creative Commons
Attribution 4.0 International (CC BY 4.0). Reuse with credit to
"Spacelike AI" and a link to https://spacelike.ai/.

## Contact

mail@spacelike.ai