Specialist-model research desk

Scorecard

What do the totals mean?

Frontier Development research deskSection 09 · Page 6 of 13Last updated 26 October 2026
Score / 100, after gates and coverage Suggested interpretation Action
80–100 Strong evidenced fit to this brief Negotiate scope and verify remaining conditions; not an automatic award
65–79.9 Conditional fit Address material weaknesses or commission a narrower pilot
Below 65 Insufficient current fit Rescope, obtain better evidence or do not proceed

These bands are decision aids. A score of 82 is not a probability, and a two-point difference should not imply false precision. Investigate which criteria drive a close result and whether plausible weight changes reverse the order.

TRACE THE EVIDENCE

Section source library

This page contains no direct external citations. These are the section’s references; they do not independently validate the desk’s proposed scoring or methods.

  1. Together AI — AI Native Cloud →www.together.ai
  2. AI Risk Management Framework — NIST →www.nist.gov
  3. Demystifying evals for AI agents — Anthropic, 9 January 2026 →www.anthropic.com
  4. GroupKFold — scikit-learn documentation →scikit-learn.org
  5. A Field Guide to Rapidly Improving AI Products — Hamel Husain →hamel.dev
  6. Llama 3.3 Community License Agreement — Meta →github.com
  7. Pricing — Together AI, snapshot checked 26 October 2026 →www.together.ai