Frontier Model Release Tracker & Elo Board

Model Release Tracker: every new Claude, GPT, Gemini, Grok, or Kimi drop gets a comparative benchmark table — not just an announcement post. The Elo Board sits underneath as the open-method ranking layer.

What a release card includes

• What shipped (model family, access tier, modalities).

• Axes we score separately: reasoning, coding, tool use, refusal quality — never one mystical number alone.

• What is still unknown (eval leakage, unpublished system prompts, rate limits).

Ranking principles

• Publish the prompt packs and scoring rubric.

• Allow reader votes as a secondary signal, never as the only signal.

• Label illustrative methodology snapshots clearly — we do not invent precise Elo scores as facts.

Update cycle

Board refresh weekly. Mid-cycle model drops get a provisional release card until the next full run. Disputes and rubric changes are logged in public changelogs.