Inference.net Schematron models: built for JSON extraction

Inference.net contributes Schematron V2 Turbo at $0.03 input and $0.15 output per 1M tokens and Schematron V2 Small at $0.05 input and $0.23 output per 1M tokens to this rating. Inference.net is a serverless inference provider that also trains small task-specific models; Schematron turns HTML into schema-conforming JSON.

With 2 models here the line-up is small, so benchmark it against a comparable model from a larger vendor before committing production traffic.

• Benchmark against a comparable model from a larger vendor before committing production traffic — line-ups this size get less independent testing.
• Start with Schematron V2 Turbo at $0.03 / $0.15 per 1M tokens; it is the cheapest entry point here.
• Check the context window against your longest input — the largest here is Schematron V2 Turbo at 125K tokens.
• Confirm support terms and deprecation policy, which smaller vendors document less thoroughly.