CURATED EVERY SUNDAY
Discover what’s next
in evaluation
Each week, we select the most noteworthy new benchmarks and datasets—and distill the evaluation methods worth carrying into your own work.
CURRENT ISSUEW4209 PICKS / 08 FIELDS
More than a paper list.
Evaluation insights you can reuse.
We look beyond what was released to examine how each evaluation defines its tasks, builds its data, verifies outcomes and reflects broader shifts in research.
01Collect
Track new work across agents, coding, multimodality, embodied AI and model evaluation.
02Select
Prioritize public work with realistic tasks, rigorous validation and genuinely novel evaluation design.
03Distill
Focus on two questions: why it matters and which design ideas are worth reusing.
02 / LATEST
Issue 10 is now available
2026.10.12 — 10.18 · 9 selected projects
AgentDatasetEmbodied AIMultimodal AgentCoding AgentMultimodalModel EvaluationTool Agent
View the latest issue ↗