Laguna XS 2.1
poolside's 33B coding-agent model — 3B parameters active at a time — compressed to NVFP4 on one desktop-sized AI computer.
Setup and speed
Full results
- —Each task is run three times; the score counts all 1,650 attempts, not 550 tasks.
- —Beats Laguna S 2.1 on the same task set (496 vs 466) despite being the smaller sibling — strongest on Inbox triage and Document extraction (both 96) and Classification (48/50). Softest on Formulas & SQL at 78, which falls to 57 on hard tasks, and Spreadsheet analysis hard at 65 with step-limit failures.
- —428 of 550 tasks were correct on all three passes. Nothing was cut off by the output limit — every miss was a wrong answer or a format/step budget issue.
- —Files round: workbook and social posts passed the machine checks (quality 1.75 and 2.75). The annual report was produced but failed length (about 3.4 pages vs an eight-to-twelve-page brief, quality 2.0). The investor deck never wrote a file before the task ended.
- —Speed probe (thinking on): first visible token median 23.2s, decode about 97 tok/s (~75 words/s). TTFT not separated from first visible content on this stack.
With thinking on, the first token arrives at 0.00s but the visible answer does not start until roughly 23.17s while the model reasons. The 15.8s typical task time is that reasoning plus the answer.
Files it produced
Four real documents from one fact sheet, produced once. Passed is machine-checked against the fact sheet and the brief; quality is a separate 1–5 score from a published rubric and never changes whether a file passed. How files are graded →
An 8–12 page annual report with six required sections: shareholder letter, financial review, category performance, customer and geographic performance, risk factors, and outlook.
Download DOCXA two-sheet workbook: a Summary comparing FY2025 to FY2024, and a Monthly sheet with all twelve months.
Download XLSXA ten-slide investor presentation following a fixed outline, with the brand palette applied and at least one real chart object.
Not produced — the model never wrote this fileAn X post of 280 characters or fewer and a LinkedIn post, both announcing the results, text only.
Download MDEvery artifact below is the file the model actually produced, published unedited so you can open it yourself. These are AI-generated documents about a fictional company — read them as samples of the model's work, not as advice or a real annual report. The workbook's formula cells hold real formulas — a viewer that does not recalculate may show them blank until Excel opens the file.