Category 10

AI Evals

Model testing, benchmarking and evaluation platforms.

Standout picks across editions

The current edition's pick, highlighted, plus earlier picks in this category from our backfilled archive — sourced from credible outlets (Hacker News, GitHub, TechCrunch and more) and checked against each product's official site.

This edition's pick · Nov 16–30, 2026
AA

Artificial Analysis

Independent AI model benchmarking

Independent model benchmarking & intelligence index.

Why it matters

Two dated releases this edition — Intelligence Index v5 (Nov 18) and an MKA agent benchmark with DeepMind (Nov 17) — cement it as a go-to independent evaluation reference.

Nov 1–15, 2026 edition
W

WebsiteBench

Can agents rebuild full-stack sites?

Benchmark: agents reverse-engineer websites via browser exploration.

WebsiteBench, from TIGER-AI-Lab, is a benchmark that tests whether AI agents can reverse-engineer a full-stack version of a website by exploring it through a browser. It spans 70 tasks and measures pass rate, compute cost, token usage and interaction turns, with a public leaderboard comparing frontier and open models. It gives buyers and researchers an independent read on how well agents can actually replicate real production web apps.

Why it mattered

A TIGER-AI-Lab benchmark testing whether agents can reverse-engineer full-stack versions of websites through browser exploration — 70 tasks with a public leaderboard.

Evidence dated 2026-11-04github ↗official ↗
Sep 16–30, 2026 edition
T

TinyAIArena

Watch AI agents battle it out

Head-to-head arena pitting AI agents against each other.

Why it mattered

A public arena where small AI agents compete head-to-head — a lightweight, watchable take on agent evaluation and comparison.

Evidence dated 2026-09-27hackernews ↗
Researched · pending approvalVisit site ↗
Sep 1–15, 2026 edition
DL

Dyno Lab

AI safety & interpretability workbench

Local alignment and interpretability workbench for Apple Silicon.

Why it mattered

A local AI safety, alignment and interpretability workbench for Apple Silicon — activations, probes and SAEs with a Python SDK and MCP.

Evidence dated 2026-09-04github ↗
Researched · pending approvalVisit site ↗

How picks are selected

Candidates are scored on market traction, recent momentum, product innovation, community interest, practical usefulness and source credibility. Only researched, fact-checked products are published — if no candidate meets the evidence threshold, we show no pick rather than forcing a weak selection.