Ineffable Intelligence

Superlearner, reinforcement learning, self-play AI

GB$5.1B privateest. 2025ineffable.ai (opens ineffable.ai)Checked 11 Oct

The brief

Analyst view
  • Record $1.1B seed round in April 2026 at $5.1B valuation, Europe's largest ever.
  • Co-led by Sequoia and Lightspeed; Nvidia, Google, UK Sovereign AI Fund participated.
  • David Silver (AlphaGo lead) + 4 DeepMind alumni. First models expected late 2026.

Technical approach

As reported
Self-play learning

Knowledge discovery without human data

Learns by generating problems internally, not from human-labeled data. Discovers knowledge human data cannot capture.

Reinforcement learning

Full RL ambition vs. LLM scaling

Uses reward signals to build understanding, not compress human knowledge. Discovers through autonomous interactions.

Goal

Exceed human knowledge boundaries

Discover theorems unproved by humans and scientific frameworks without human language.

Primer