← DeepSeek V4 Pro · API · Max effort
Evidencesupersededvendor_reported
Humanity's Last Exam
DeepSeek V4-Pro-Max — HLE
ev_2026_07_dsv4pro_hle
Result
37.7%
Confidence B
- Tested
- 2026-04-24
- Runs
- 1
Configuration
How it was measured
- Exact model
- deepseek-v4-pro
- Reasoning
- max
- Agent harness
- none
- Tool access
- none (closed-book)
- Prompting
- vendor HLE
SUPERSEDED by independent AA HLE (ev_2026_07_dsv4pro_aa_hle). Kept for audit trail.
Caveats
Known limitations
- Vendor-reported closed-book HLE.
- Superseded by independent AA float for scoring.
Weighting
Evidence weight factors
The engine weights each evidence row by these factors (0–1) when fusing scores — higher is more trusted.
relevance0.95
quality0.55
recency0.90
health0.90
comparability0.85
reliability0.70
Benchmark
Humanity's Last Exam
Frontier closed-ended academic benchmark across many subjects; often reported with and without tools.
Benchmark source →