Model results
TextQuests
Choose a shared evaluation protocol to compare model scores. Versions, sources and configurations stay separate.
Atlas snapshot captured 10/19/2026, 10:20:28 AM UTC. This is an import timestamp, not the evaluation date. Coverage may be incomplete.
TextQuests
- Source
- Center for AI Safety (reviewed publisher results)
- Version
- dashboard-unversioned
- Measurement date
- Not reported
- Metric
- % game progress · Higher is better
- Evaluation scope
- Release-level identity supported by exact named release and publisher metadata; dashboard releaseDate=2025-12-01; identity reference=https://api-docs.deepseek.com/news/news251201. Exact V3.2 release, matching date and official V3.2 announcement distinguish stable release from V3.2 Exp. Thinking remains configuration metadata. Dated provider endpoint revision is not reported. CAIS dashboard TextQuests progress field across the 25-game benchmark; publisher configuration=deepseek-v3.2-thinking; clue setting, step budget, run count and executed harness revision not reported in dashboard result metadata; do not merge with the separate textquests.ai no-clues or with-clues cohorts; not completion rate or harm. result date not supplied; captured 2026-10-19.
Fewer than two comparable models. No ranking is assigned. Missing metadata is shown as unreported; matching imported metadata is not independent verification of the protocol.
DeepSeek V3.2
deepseek/deepseek-v3.2-20251201
21.2% game progress
View modelEvidence independent