Foundation + Eval Harness
Completed: July 2026
Build the basic research infrastructure that can measure model behavior reliably before changing it.
Open Model Lab
Open Model Lab Stages
Research stages, current target dates, and completed evidence.
Research stages
SFT Pipeline + Data Quality
Target: November 2026
Measure the behavioral difference between a base model and an instruction-tuned model.
Preference Optimization / DPO
Target: December 2026
Use preference data to improve SFT behavior in a more controlled way, then measure behavioral side effects.
Agent Harness / Tool Use
Target: January 2027
Move the model from passive question-answering into a tool-using agent setup.
Agent Evals + Long-Horizon Tasks
Target: February 2027
Measure agent success and failure taxonomy on multi-step tasks.
Safety / Red Teaming / Refusal Quality
Target: March 2027
Measure safety behavior by quality and balance, not just refusal rate.
Reasoning Behavior + Process Evaluation
Target: April 2027
Evaluate not only final answers, but where reasoning processes break down.
Monitorability / Interpretability Start
Target: May 2027
Start analyzing model behavior with internal signals in addition to external scores.
Multimodal / UI Understanding / Computer Use
Target: June 2027
Enter multimodal model work through screenshot and UI-understanding tasks.
Training / Inference Systems Efficiency
Target: July 2027
Add systems and profiling knowledge to make research experiments more efficient.
Data Efficiency + Scaling Ladder
Target: August 2027
Measure whether better data mixtures produce better behavior with the same compute.
Final Integration + Public Portfolio
Target: September 2027
Turn the 12-stage work into a showable, reproducible, publishable portfolio.