The discriminative, contamination-resistant, fully transparent LLM ranking — updated live.
-
Wed, Aug 5, 2026 4
-
docs: methodology white paper (17-page LaTeX + PDF + BibTeX)
…
Full technical white paper: benchmark failure analysis, design principles, 2PL IRT and Bradley-Terry mathematics, robustness metrics, score structure and interpretation guide, governance, limitations, comparison table, roadmap, 33 academic references. Compiles with latexmk -pdf. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
-
fix(worker): cap completion budget by model context window
…
max_tokens=16384 caused HTTP 400 on small-context models (e.g. 16k-window models); budget is now min(16384, context_length - 4096) with a 2048 floor. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
-
feat(web): deep-detail methodology, privacy policy, terms of service, live efficiency frontier
…
- Methodology: scoring mathematics (2PL MAP, Fisher SE, BT MM + SE, composite formulas), item lifecycle & bank hygiene, full judge protocol, known limitations, versioning/audit section, contact block - New /privacy and /terms pages (contact: Simon-Pierre Boucher, contact@spboucher.ai), footer links - Efficiency frontier + hero stats now poll live (/api/v1/efficiency) - Rankings display current INDEX_VERSION only (old runs remain as immutable audit history) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
-
feat: LLM Index platform v0.2.0 — discriminative, contamination-resistant live LLM ranking
…
IRT 2PL + Bradley-Terry scoring over 12 domains with dynamically generated items: hardened math/reasoning/code/knowledge/multilingual/instruction, home-made simulated agentic (+context-load) and terminal benches, SVG logo duels (3-judge cross-provider panel), vision OCR. Robust extraction cascade, parallel evaluation lanes with live-updating leaderboard, per-model full transparency pages, public API v1, detailed methodology with references. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
-