← All models

Think-Reason-Learn Ensemble on VCBench

Think-Reason-Learn Ensemble scores F0.5 37.9 on VCBench, the venture capital benchmark from the University of Oxford and Vela Research, with 40.6% precision and 30.1% recall on the 4,500-founder private test set. That is rank 1 of 34, 3.5× the F0.5 of tier-1 VCs and 4.4× Y Combinator, at $0.87 per 1,000 founders scored.

Think-Reason-Learn Ensemble on the VCBench leaderboard today
Rank
#1 of 34
F0.5
37.9
Precision
40.6%
Recall
30.1%
Cost / 1k founders
$0.87

Scored on the 4,500-founder private test set, mean over three folds. That is 3.5× the F0.5 of tier-1 VCs. See the full leaderboard.

Compared with the reference rows

EntryPrecisionRecallF0.5Cost / 1k
Think-Reason-Learn Ensemble40.6%30.1%37.9$0.87
Tier-1 VCs23.0%5.2%10.7n/a
Y Combinator14.0%6.9%8.6n/a
Random Classifier9.0%9.0%9.0$0

Human rows are normalized to the dataset's 9% success rate, so every entry is compared on the same base rate. Costs use list prices on 2026-09-24.

How it was scored

A mix of families: LLM-written rules and classical models are combined into one score, so an LLM still runs at scoring time. Think-Reason-Learn Ensemble used about 20,800 input and 7,448 output tokens per founder on Typesafe System One (jev-1.13.0), measured from the API usage fields. Cost is tokens times the provider's list price on 2026-09-24 ($0.042 input and $0 output per million tokens), so it recomputes when prices change. Build costs such as question or policy generation are excluded.

Read more

Frequently asked questions

What does Think-Reason-Learn Ensemble score on VCBench?
Think-Reason-Learn Ensemble scores F0.5 37.9 on VCBench, with 40.6% precision and 30.1% recall, rank 1 of 34 as of 2026-10-26.
Does Think-Reason-Learn Ensemble beat human investors at predicting founder success?
Yes. Its F0.5 is 3.5 times that of tier-1 VCs (10.7) and 4.4 times Y Combinator (8.6), after both are normalized to the dataset's 9% base rate.
How much does Think-Reason-Learn Ensemble cost per 1,000 founders?
About $0.87 per 1,000 founders at list prices on 2026-09-24. A mix of families: LLM-written rules and classical models are combined into one score, so an LLM still runs at scoring time. Think-Reason-Learn Ensemble used about 20,800 input and 7,448 output tokens per founder on Typesafe System One (jev-1.13.0), measured from the API usage fields. Cost is tokens times the provider's list price on 2026-09-24 ($0.042 input and $0 output per million tokens), so it recomputes when prices change. Build costs such as question or policy generation are excluded.