Fetching from the wire…
Research2026-06-07 · source-backed
ALE is 1,000+ economically meaningful tasks mapped to the US occupational taxonomy across finance, healthcare, legal, engineering, and manufacturing. The hardest tier averages a 2.6% full pass rate. Source: Agents' Last Exam The thesis is explicit: SWE-bench made coding measurable and improvable, ALE wants to do that for the rest of GDP-weighted knowledge work. Pair it with SWE-Marathon and the message is consistent. We measure the easy stuff well and the hard stuff barely at all.
Each link below shares sources, entities, or timing with this story.
Codex benchmarked against ALE / Shared entities / Shared topic / What happened next / Tension
Linked by a graph relationship (Codex benchmarked against ALE); both cover ALE, Last Exam; overlapping topics (agent, exam, tier).
Codex benchmarked against ALE / Shared entities / Shared topic / What happened next
Linked by a graph relationship (Codex benchmarked against ALE); both cover Last Exam, SWE; overlapping topics (agent, exam).
Codex benchmarked against ALE / Shared entity: SWE / Shared topic / Earlier coverage
Linked by a graph relationship (Codex benchmarked against ALE); both cover SWE; overlapping topics (agent, coding).
Linked by a graph relationship (Codex benchmarked against ALE); both cover SWE; overlapping topics (agent, full).
Linked by a graph relationship (Codex benchmarked against ALE); both cover SWE; overlapping topics (agent, coding).
Shared entities / Shared topic / What happened next
Both cover Last Exam, SWE; overlapping topics (coding, exam); picks up the Last Exam thread on 2026-07-31.
Codex benchmarked against ALE / Shared entity: SWE / What happened next / Tension
Linked by a graph relationship (Codex benchmarked against ALE); both cover SWE; picks up the SWE thread on 2026-08-06.
Codex benchmarked against ALE / Shared entity: SWE / Earlier coverage / Tension
Linked by a graph relationship (Codex benchmarked against ALE); both cover SWE; earlier SWE coverage from 2026-04-25.