Benchmark recording · 29 September 2026

An AI agent works a WebAssign-style calculus homework

One recorded run on a benchmark replica modeled on Cengage WebAssign, sped up 10×. The replica is a test copy we built, not the real site.

16 of 16graded parts correctSubmitted

Recorded 29 September 2026 · 10× speed · 8 min 6 s of real time, shown in a 56 s clip

What happens in the clip

The agent works a WebAssign-style calculus homework on volumes of solids: it picks graphs and 3D figures as answer choices, opens each part of a long accordion question, types answers in math notation, and submits.

Step by step

  1. Works a WebAssign-style calculus homework on volumes of solids, on the replica.
  2. Picks graphs and 3D figures as answer choices.
  3. Opens each part of a long accordion question in turn.
  4. Types answers in math notation.
  5. Submits the homework.
  6. Replica read back: 16 of 16 graded parts correct.

What this replica tests

Graphs as answer choices, 3D figures, a 10-part accordion, math notation. The side panel in the video shows how many graded parts are complete and whether the assignment has been turned in.

About this recording

This recording is a separate run from the scored benchmark run on 22 September 2026, and that run got the same result on this replica: 16 of 16, submitted.

A benchmark replica we built for testing, modeled on Cengage WebAssign. It is not the real site, and the platform's maker is not affiliated with it and has not endorsed it. The agent is the Base plan's vision engine, run in a standalone test harness, not the full desktop app. The on-screen label “Luke’s Empire” is the benchmark’s internal name.

This clip is one of seven on the benchmark page, which has the full method and the scored results. It is also shown on the Cengage and WebAssign pages.

Other recorded runs

Silent Student is the desktop app this engine runs in. See what it does on WebAssign.

Updated 19 October 2026