Sites that watch the AI field itself and publish a continuously refreshed number from that watching: release timelines, independent benchmarks, open datasets, composite rankings, preference arenas, and usage statistics. The category’s dividing line is what the number measures, since each member answers a different question about the same field.
- AI Release Tracker - a free, single-operator timeline of frontier releases since ChatGPT, with launch-day scores and flat-file exports for agents.
- Artificial Analysis - independent quality, price, and speed measurement across 718 models, 100+ providers, and the chips underneath.
- BenchLM - a solo-built composite re-ranking 1,004 tracked models into one evidence-labeled score, with its own failed validation checks published.
- Benchmark Radar - an open benchmark catalog and search engine crawling 40 sources daily, with every record’s evidence linked, the category’s only member whose subject is the benchmarks themselves.
- bestvaluemodel - a solo-built, six-hourly refreshed frontier view plotting Artificial Analysis’s Intelligence Index against API price, with a budget lookup table and a per-run diff feed.
- Epoch AI - the research nonprofit’s open datasets and trend series on models, compute, data centers, and AI’s trajectory.
- LiveBench - a contamination-limiting public benchmark that rotates questions and scores by objective ground truth, with questions, answers, and judgments downloadable.
- LLM Stats - a conservative composite leaderboard over 408 models with the category’s only MCP server for agents.
- LMArena - blind human-vote arenas ranking models by preference, the field’s most cited and most gamed signal.
- Margin Lab - daily degradation trackers running Claude Code and Codex against a fixed SWE-Bench Pro subset with 95% confidence intervals, plus an open-source eval runtime.
- OpenRouter Rankings - real token spend through the OpenRouter gateway, CC BY licensed, the category’s only revealed-preference number.
- Vectara Hallucination Leaderboard - Vectara’s HHEM-judged ranking of how often models hallucinate when summarizing short documents, with an open variant of the judge model downloadable.
- whatmodel.sh - a curl-first composite that fits public results onto one IRT scale and filters to what your machine can run, afford, and download raw.
Its members are compared on shared rows in the Trackers and Leaderboards Feature Matrix.
Changes #
- 2026-09-24 - Added AI Release Tracker.
- 2026-09-24 - Added Artificial Analysis.
- 2026-09-24 - Added Epoch AI.
- 2026-09-24 - Added LLM Stats.
- 2026-09-24 - Added LMArena.
- 2026-09-24 - Added OpenRouter Rankings.
- 2026-10-06 - Added LiveBench.
- 2026-10-10 - Added Benchmark Radar.
- 2026-10-11 - Added BenchLM.
- 2026-10-11 - Added Margin Lab.
- 2026-10-13 - Added Vectara Hallucination Leaderboard.
- 2026-10-17 - Added whatmodel.sh.
- 2026-10-25 - Added bestvaluemodel.