# Evaluation Becomes a Standalone Software Market

Model evaluation, grading and benchmarking separate from the labs into an independent tooling and services market, with cheap open-weight judges collapsing the unit cost of grading; expect recurring vendor launches, funding rounds and consolidation in eval infrastructure as buyers demand verification they do not have to take from the model vendor.

- Conviction: 38 / 100 (forming)
- Horizon: Emerging (watchlist)
- Tracking since: 2026-08-04T00:00:00.000Z
- Last updated: 2026-08-04T06:16:49.912Z
- Canonical: https://polylog.news/ai/trends/ai-evaluation-tooling-market
- Publisher: Polylog
- Affected regions: United States

## Recent evidence

- [confirms] AI Evaluation Turns Into a Product Market as Researchers Show Cheap Open Models Can Grade Proofs (2026-08-04): A published forecast sizes model evaluation and benchmarking tools at roughly $9.6 billion by 2035 against about $1.15 billion last year, and new research shows cheap open-weight models can grade proofs in place of expensive frontier graders. Both the demand-side forecast and the supply-side cost collapse point at an independent eval vendor layer.