FOTOhub raises $5.1M (~20.4M PLN) to develop proprietary AI models and fuel international expansion
FOTOhub has closed a $5.1M seed round (~20.4M PLN) with 3 strategic investors. The capital will fund development of proprietary AI models - FOTOcore, GABRIEL, IDA Creative, and Laura Coder - alongside expansion through Google for Startups and AWS Startup programs.
FOTOhub raises $5.1M to develop proprietary AI models
Bydgoszcz, Poland - March 5, 2026 - FOTOhub, a Polish creative-tech AI platform, has officially closed a seed funding round of $5.1 million USD (approximately 20.4 million PLN), bringing together three strategic investors into the capital of FOTOhub Polska Sp. z o.o. and FOTOhub Investment. The funds will be primarily allocated to developing proprietary AI models and GPU infrastructure.
What is FOTOhub
FOTOhub is a full-stack AI creative platform built for photographers, agencies, and content creators. We combine image generation, video production, audio synthesis, AI automation, and community features in one place.
FOTOhub is not a wrapper on top of a single API - it's a complete application ecosystem with its own orchestration layer, custom AI models, multi-provider video rendering, real-time collaboration, marketplace, and a professional video editor with server-side rendering.
Round details
| Parameter | Value |
| Round type | Seed |
| Amount | $5.1M (~20.4M PLN) |
| Investors | 3 strategic investors |
| Closing date | March 5, 2026 |
| Structure | FOTOhub Polska Sp. z o.o. + FOTOhub Investment |
Proprietary AI models
We develop, fine-tune, and run our own AI model stack. Models are trained on proprietary datasets, served on Google Cloud GPU infrastructure, and exposed to users through a unified API.
FOTOcore AI (v0.2) - orchestration and routing
Central orchestration and routing layer. Analyzes incoming requests (modality, quality target, cost budget, user tier) and routes them to the right model pipeline. Handles load balancing, automatic failover, retry logic, and credit-cost optimization across all FOTOhub models and partner APIs.
- Architecture: stateless orchestrator on Deno runtime with provider-specific adapters, response normalization, and unified error handling
- Routing latency: <50ms (p95)
- Throughput: 10,000+ requests/min
- Managed models: 4 internal + 5 external (API)
- SLA: 99.95% uptime
FOTOcore AI also powers the Flows FH automation system, enabling users to chain AI operations into autonomous sequences.
GABRIEL (v0.4) - image generation and editing
Vision model for image generation, editing, and composition. Fine-tuned on proprietary photography and creative datasets using LoRA adapters. Trained and served on Google Cloud GPU infrastructure (Vertex AI).
- Capabilities: text-to-image, image-to-image, multi-reference composition (up to 13 input images), inpainting, outpainting, background swap, style transfer, 4K upscaling, brand-consistent series generation
- Architecture: multi-checkpoint pipeline - routing between specialized checkpoints depending on task type (photorealism, illustration, product, fashion)
- Brand AI: users can train their own GABRIEL LoRA variants for brand-specific generation
- Models: FOTOcore 4.0 Ultra, FOTOcore 4.0, FOTOcore 4.0 Fast, FOTOcore 3.0 HD, FOTOcore 3.0 Fast
- Pre/post-processing: image normalization, resize, mask preparation, quality scoring, NSFW filtering, metadata enrichment
IDA Creative (v0.3) - audio and voice
Audio and voice model for synthesis, cloning, transcription, and audio production. Custom voice models trained per-user on Google Cloud GPU. Audio processing pipeline with proprietary normalization, LUFS targeting, and multi-format output.
- Music generation: 12+ genres with BPM and key control
- Speech synthesis (TTS): 16+ languages, 10+ voice profiles
- Voice cloning: instant and professional modes
- Speech-to-speech: real-time voice conversion
- Transcription: with speaker diarization and emotion detection
- Sound effects: SFX synthesis with variations
- Audio mastering: normalization, EQ, multiband compression
- Stem separation: 2/4/5-way decomposition
- Inference: Vertex AI with auto-scaling, output in MP3, WAV, FLAC, OGG
Laura Coder (v0.1-alpha) - code generation assistant
Model in active development for platform automation. Targeted for custom integration scripting, workflow automation code, and API client generation.
- Status: active development (R&D)
- Languages: TypeScript, Python, SQL, YAML
- Context: 128K tokens
- Planned beta: Q3 2026
Platform at a glance
| Module | Description |
| AI Studio | 17+ generative models, text-to-image, editing, style transfer, multi-reference composition |
| Video Studio | video generation (Veo 3.1, Kling, Wan), professional editor with server-side rendering |
| Audio Studio | speech synthesis, music, SFX, mastering, stem separation, voice cloning |
| Flows FH | AI workflow automation, chaining operations into sequences |
| Canvas | AI-powered graphic editor, drawing board, mask editor with SAM |
| API | 52 REST endpoints, full documentation, SDK |
Key metrics
- 460,000+ registered users
- 11,000,000+ AI generations
- 95+ AI applications in the ecosystem
- 17+ generative models available
- 4 proprietary AI models (FOTOcore, GABRIEL, IDA Creative, Laura Coder)
- TOP 2 Crunchbase Poland in AI category
- top 1% of global Crunchbase index
GPU and compute infrastructure
FOTOhub operates on dedicated GPU infrastructure provided by Google Cloud Platform. We have our own GPU processors allocated for training and serving AI models, giving us full control over performance, costs, and response times.
- GPU processors - dedicated NVIDIA units on Google Cloud for training and serving generative models
- 3rd generation SSD storage - fast NVMe drives for handling large training datasets and model caching, ensuring minimal latency on data reads
- Infrastructure investment - 40% of raised capital ($2.04M) allocated to GPU cluster expansion and compute scaling
- Model training - proprietary training pipelines on Vertex AI with auto-scaling, enabling fine-tuning of GABRIEL and IDA Creative models on proprietary datasets
- Auto-scaling - dynamic GPU resource scaling based on load, cost optimization during off-peak hours
- Multi-region - distributed infrastructure across Google Cloud regions (europe-west, us-central) ensuring low latency for global users
- Redundancy - multi-cloud strategy (Google Cloud + AWS) guaranteeing 99.95% availability and eliminating single-vendor dependency
With our own GPU infrastructure, FOTOhub can train and update AI models without relying on external API providers, resulting in lower latency, better costs, and full control over the quality of generated content.
Accelerator programs
Alongside closing the round, FOTOhub has been accepted into two global accelerator programs:
Google for Startups
- Google Cloud Credits for Vertex AI, Cloud Run, and infrastructure
- priority access to Gemini, Imagen, and Veo models
- dedicated technical advisor and business mentorship
- membership in a global founder and investor network
FOTOhub leverages Google models as key components - Imagen 4.0 as the foundation for GABRIEL, Veo 3.1 in Video Studio.
AWS Startup
- AWS Credits for EC2, S3, Lambda, and SageMaker
- Business Support Plan and architectural consultations
- multi-cloud strategy: Google Cloud (AI) + AWS (redundancy)
- 99.95% SLA guarantee and single-vendor dependency elimination
Pricing plans
FOTOhub offers 7 subscription tiers - from free to Enterprise:
| Plan | Price | Credits/mo |
| Free | $0 | 50 |
| Starter | ~$8 | 200 |
| Medium | ~$22 | 1,500 |
| Professional | ~$65 | 5,000 |
| Business | ~$175 | 20,000 |
| Team | $99 | team collaboration |
| Enterprise | custom | unlimited |
Every plan includes unlimited storage with automatic format optimization and global CDN distribution.
Use of proceeds
- 40% - GPU infrastructure and compute cluster expansion
- 30% - R&D, continued development of FOTOcore, GABRIEL, IDA Creative, and Laura Coder models
- 20% - international expansion (DACH, UK, US)
- 10% - team and operations
The funding, combined with Google and AWS partnerships, positions FOTOhub as the leader of the Polish AI creative-tech market. A planned Series A round (50M PLN) in Q4 2026/Q1 2027 will enable further infrastructure scaling and international market entry.
About FOTOhub
FOTOhub is a Polish AI platform founded in September 2025. We combine image generation, video production, audio synthesis, and AI automation in a single ecosystem with proprietary AI models, a professional video editor with server-side rendering, and a marketplace for creators. The only Polish creative-tech company in the TOP 2 Crunchbase Poland and top 1% of the global index.
Press contact: [email protected]
Investor contact: [email protected]
Website: fotohub.pl