Archive — US — techcrunch
Vals, backed by Andreessen Horowitz, is looking to become the gold standard forAI benchmarking
Vals, a startup backed by Andreessen Horowitz, aims to set a new standard for AI benchmarking, addressing outdated systems. Founded in 2024, it has rapidly gained traction, raising $40M recently. Co-founder Rayan Krishnan emphasizes the need for benchmarks that reflect modern model capabilities in industries like law and finance, highlighting the importance of accurate evaluation for improving AI systems and establishing trust as AI models become integral to the economy.
Key points
- Vals formed in 2024 to improve AI benchmarking; current benchmarks are outdated.
- Company raised $40M in Series A led by Andreessen Horowitz after previous funding rounds.
- Co-founder Rayan Krishnan highlighted that Vals measures AI's performance on industry-specific tasks, not just general knowledge.
- Instead of public tests, Vals evaluates real-world impacts and negative implications of AI models.
- Company revenues have increased eightfold year-over-year; staffing increased from 8 to 25 and is set to grow further.
- The startup's evaluations are becoming crucial for companies acquiring new AI models and for public trust in the sector.
Why it matters
- In 2025, AI continues to integrate into various sectors, necessitating robust evaluation methods.
- The trend of AI companies going public creates demand for reliable benchmarking as part of their financial disclosures.
- Historical context: AI benchmarks have long been challenged for relevance, reflecting a broader issue in tech evaluation standards.
Vals is positioned to revolutionize AI benchmarking, providing essential evaluations for growing industries and fostering trust in AI as it becomes a central economic component.