Agents' Last Exam
by Snorkel AI · Benchmark · Text & code
Long-horizon professional workflow benchmark where GPT 6.2 Astra scores above Claude Fable 5.2
Evidence
- Sep 24Opus 5.5 not yet on the leaderboard at recording
by Snorkel AI · Benchmark · Text & code
Long-horizon professional workflow benchmark where GPT 6.2 Astra scores above Claude Fable 5.2