GemVC-v0 on VCBench
GemVC-v0 scores F0.5 32.9 on VCBench, the venture capital benchmark from the University of Oxford and Vela Research, with 39.4% precision and 20.3% recall on the 4,500-founder private test set. That is rank 5 of 32, 3.1× the F0.5 of tier-1 VCs and 3.8× Y Combinator, at $2.87 per 1,000 founders scored.
- Rank
- #5 of 32
- F0.5
- 32.9
- Precision
- 39.4%
- Recall
- 20.3%
- Cost / 1k founders
- $2.87
Scored on the 4,500-founder private test set, mean over three folds. That is 3.1× the F0.5 of tier-1 VCs. See the full leaderboard.
Compared with the reference rows
| Entry | Precision | Recall | F0.5 | Cost / 1k |
|---|---|---|---|---|
| GemVC-v0 | 39.4% | 20.3% | 32.9 | $2.87 |
| Think-Reason-Learn Ensemble | 40.6% | 30.1% | 37.9 | $0.87 |
| Tier-1 VCs | 23.0% | 5.2% | 10.7 | n/a |
| Y Combinator | 14.0% | 6.9% | 8.6 | n/a |
| Random Classifier | 9.0% | 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
An LLM reads each founder profile at scoring time and returns a prediction. GemVC-v0 used about 284 input and 1,113 output tokens per founder on gemini-2.5-flash, estimated from the prompt and the saved responses; reasoning models are assumed to think about 1,000 tokens per founder. Cost is tokens times the provider's list price on 2026-09-24 ($0.3 input and $2.5 output per million tokens), so it recomputes when prices change. Build costs such as question or policy generation are excluded.
Frequently asked questions
- What does GemVC-v0 score on VCBench?
- GemVC-v0 scores F0.5 32.9 on VCBench, with 39.4% precision and 20.3% recall, rank 5 of 32 as of 2026-10-07.
- Does GemVC-v0 beat human investors at predicting founder success?
- Yes. Its F0.5 is 3.1 times that of tier-1 VCs (10.7) and 3.8 times Y Combinator (8.6), after both are normalized to the dataset's 9% base rate.
- How much does GemVC-v0 cost per 1,000 founders?
- About $2.87 per 1,000 founders at list prices on 2026-09-24. An LLM reads each founder profile at scoring time and returns a prediction. GemVC-v0 used about 284 input and 1,113 output tokens per founder on gemini-2.5-flash, estimated from the prompt and the saved responses; reasoning models are assumed to think about 1,000 tokens per founder. Cost is tokens times the provider's list price on 2026-09-24 ($0.3 input and $2.5 output per million tokens), so it recomputes when prices change. Build costs such as question or policy generation are excluded.