Snorkel expands Open Benchmarks Grants 10x to a $30M commitment
windx0303 · x · 2026-10-09
Snorkel AI announced a 10x expansion of its Open Benchmarks Grants to a $30M commitment to support open frontier AI evaluation.
- Funding new benchmarks: safety and alignment, cybersecurity, physical AI, long-running agents, open-ended outputs, and dynamic settings
- Open Benchmarks Red Team: continuously finding and fixing weaknesses like reward-hacking, contamination, and broken verifiers
- Snorkel Research Fellowship: supporting independent evaluation researchers with compute and engineering support
- Its earlier $3M phase partnered with Terminal-Bench, TB-Science, ARC-AGI-3, OSWorld 2.0, and Senior SWE-Bench teams; rolling applications are open
Related event: Snorkel Expands Open Benchmarks Grants Tenfold to $30 Million(2 posts)→
More from Venture
- Catalyst raises $30M for autonomous AI trading agent, hiring engineers — jzlegion · 2026-10-09
- OpenAI tells investors annualized revenue hit $50B in September, short of reported $70B — deedydas · 2026-10-09
- Shopify adds Grok and agent connectors; sellers may build AI storefronts to dodge 2.9% fees — Daniel_Farinax · 2026-10-09
- 68% of searches end without a click: how startups can win AI search recommendations — nikvassev · 2026-10-09
- Ferryman: cross-posting tool targeting $10,000 MRR, $30-100/mo tiers — KevinNaughtonJr · 2026-10-09
- Investor bets Anthropic's $65BN ARR is inflated, expects $45BN or less after write-downs — MatthewChang · 2026-10-09