Research Division — Reasoning & RL
Verifiable Test-Time Compute & Deep Reasoning
How Alchemax Labs bridges the gap between statistical pattern matching and formal, multi-step logical deduction across mathematics, code synthesis, and scientific discovery.
Pillar 01 · Verification Loops
Reinforcement Learning with Formal Feedback
Instead of relying solely on human preference rankings (RLHF), which can reward confident-sounding falsehoods, our o-series models train directly against compilers, Lean 4 proof assistants, and symbolic math kernels.
Pillar 02 · Search & Backtracking
Latent Tree Search at Inference Time
When solving novel problems, Alchemax-o3 Pro dynamically allocates compute to explore competing hypotheses in parallel, verifying intermediate invariants before committing to a solution path.