Flinky
Get Started

The roadmap.

Where Flinky is headed. Here's what we've shipped, what we're building right now, and what's coming next.

Shipped In progress Planned
  1. Shipped Shipped

    One-click private models

    Launch your own private instance of open-source models like Qwen, GPT-OSS and Kimi from the dashboard no setup required.

  2. Shipped Shipped

    Sovereign EU infrastructure

    Inference runs on MeluXina, an ISO 27001-certified European supercomputer in Luxembourg. Your data never leaves Europe.

  3. Now In Progress

    Expanding the model library

    Adding more open-source models continuously so you always have access to the latest and best the community ships.

  4. Next In Progress

    Programmatic API access

    Call your private model from your own apps with a simple, OpenAI-compatible API beyond the in-dashboard chat.

  5. Next In Progress

    MeluXina-AI acceleration

    Tapping into MeluXina-AI the new EuroHPC AI Factory expansion in Luxembourg, purpose-built for AI workloads. More headroom for bigger models and faster inference, fully EU-sovereign.

  6. Now In Progress

    Autoscaling

    Automatic scaling for heavier production workloads, so capacity follows demand instead of each private model running on fixed dedicated capacity.

  7. Planned Planned

    Team workspaces

    Shared private models and usage controls for your whole organisation, with roles and access management.

// Models

50+ open models, ready to launch

Model runs privately on MeluXina. New ones are added continuously search to find one.

Available (62) Coming soon
  • Kimi K2.5 1T
  • Kimi K2.6 1T
  • Qwen3.5-397B-A17B 397B
  • Qwen3.5-397B-A17B (Int4) 397B
  • MiniMax-M2.5 229B
  • Mistral-Medium-3.5 128B
  • Qwen3.5-122B-A10B 122B
  • NVIDIA Nemotron-3-Super-120B 120B
  • GPT-OSS 120B 120B
  • Qwen3-Next-80B-A3B 80B
  • Apertus-70B-Instruct 70B
  • Hypernova 60B 60B
  • Qwen3.5-35B-A3B 35B
  • Qwen3.6-35B-A3B 35B
  • Deepseek Coder 33B 33B
  • Qwen2.5-32B-Instruct 32B
  • Qwen3-32B 32B
  • Gemma 4 31B 31B
  • NVIDIA Nemotron-3-Nano-30B 30B
  • Qwen3-30B-A3B-Instruct 30B
  • Qwen3-Coder-30B-A3B 30B
  • Gemma 3 27B 27B
  • MedGemma 27B 27B
  • TranslateGemma 27B 27B
  • Qwen3.5-27B 27B
  • Gemma 4 26B-A4B 26B
  • Mistral-Small-3.2-24B 24B
  • GPT-OSS 20B 20B
  • Ministral-3-14B-Instruct 14B
  • Gemma 2 9B 9B
  • Qwen3-8B 8B
  • Apertus-8B-Instruct 8B
  • Mistral-7B-Instruct-v0.3 7B
  • Teuken-7B v0.6 7B
  • Qwen2.5-7B-Instruct 7B
  • Qwen1.5-4B-Chat 4B
  • Phi-4-mini-instruct 3.8B
  • Qwen3-0.6B 0.6B
  • InternVL3.5-241B-A28B 241B
  • Qwen3-VL-235B-A22B 235B
  • InternVL3.5-38B 38B
  • Qwen3-VL-32B 32B
  • InternVL3.5-30B-A3B 30B
  • Qwen3-VL-30B-A3B 30B
  • Qwen3-VL-8B 8B
  • DeepSeek OCR 2 3B
  • LightOnOCR-2 1B
  • PaddleOCR-VL 1.5 0.9B
  • GLM-OCR 0.9B
  • Granite Docling 258M
  • Qwen-Image 20B
  • Z-Image-Turbo 6B
  • FLUX.2-klein-4B 4B
  • Qwen-Image-Edit-2509 14B
  • Wan 2.2 T2V 14B
  • Wan 2.2 TI2V 5B
  • LTX-2 Video
  • Wan 2.2 I2V 14B
  • Wan 2.2 S2V 14B
  • Qwen3-ASR-1.7B 1.7B
  • Qwen3-ASR-0.6B 0.6B
  • Octen Embedding 8B
// Shape what's next

Want to influence the roadmap?

Beta users get a direct line to the team and early access to everything above.

Get Started