Real runs, unedited
Two jobs run on the production engine exactly as a customer would run them. The reports below are the agent's own summary.md; figures are the PNGs it saved. Input datasets are synthetic evaluation sets.
QC of a CRO IC50 deliverable (NCI-N87, two ADC candidates + benchmark + control)
class specialist · standard model · 11 min wall time · 34 tool calls · 8 artifacts · 173k tokens
Prompt: “Take this data from the CRO and QC it and return a summary of the results.”
Inputs: 20260531_STUDY-01_2D cell proliferation assay_NCI-N87_IC50.xlsx
Proteins of unknown function in extremophiles: the MEMO1 family across the three domains of life
class research · frontier model · 39 min wall time · 90 tool calls · 48 artifacts · 1.11M tokens
Prompt: “Give me an example of some proteins of unknown function in extremophiles. Choose the most interesting, and build a phylogenetic tree with related sequences across all domains of life, trace the evolution of key catalytic residues, and infer ancestral functions. In your final summary, don't include markdown links to raw data files larger than ~1 MB — reference them by name instead.”
AI-generated analysis. Outputs may contain errors; verify numbers against the execution log and source files before relying on them. Not a substitute for review by a qualified scientist, and not medical advice: any use affecting patient care or healthcare decisions requires review by a qualified professional.