Noesion Radarnoesion.ai
Research2026-07-24

Independent testing firm Artificial Analysis reports that Anthropic's newly released Claude Opus 5 tops its benchmark for multi-step office work such as research reports, presentations and spreadsheets, at about 20% less cost per task than the previous leader. Gains come mainly from accuracy and analysis rather than the look of the output, where a rival GPT model still scores higher. The strongest settings take over 25 minutes per task.

What changed

Claude Fable 5 led the benchmark at $22.30 per task.

What it unlocks

Running long multi-file research, report and spreadsheet tasks at a lower cost per task than the previous best-scoring model.

  • 1720 vs 1574 Elo (Fable 5)
  • $17.79 vs $22.30 per task
  • high effort: $10.41 per task
  • 36.2 min per task at max effort

What you need to act on it

  • API or product access to Claude Opus 5
  • willingness to wait 25+ minutes per task

Sources

Send this to someone who needs it

Shares the story and its sources. Nothing about you.

What does this mean for your job?

This is the story as everyone gets it. Once a week we send you the version written for your role — what changed, why it matters for the work you actually do, and one thing to try. Free while we tune it.