@modelschemas/client RateCard type does not fit evaluateCard from @modelschemas/rate-card #494

Problem

The RateCard from @modelschemas/client does not fit evaluateCard from @modelschemas/rate-card. Both are 0.2.1. The client's type is generated from OpenAPI (price: unknown, tables: { [key: string]: unknown }). The evaluator's type has price: RateCardPrice and tables: Record<string, Table>.

import { createModelschemasClient, getModel } from '@modelschemas/client'
import { evaluateCard } from '@modelschemas/rate-card/evaluator'

const client = createModelschemasClient({ baseUrl: 'https://modelschemas.com' })
const { data } = await getModel({ client, path: { provider: 'anthropic', modelId: 'claude-sonnet-5-5' } })
if (data?.pricing && 'price' in data.pricing) {
  evaluateCard(data.pricing, { usage: { input_tokens: 1000, output_tokens: 100 } })
}

error TS2345: Argument of type '.../@modelschemas/client/dist/generated/types.gen").RateCard'
is not assignable to parameter of type '.../@modelschemas/rate-card/dist/card").RateCard'.

So a caller must cast, or parse every card with rateCardSchema (which needs zod), to price a card that the client just returned. TanStack AI's docs page (TanStack/ai#1308, docs/advanced/model-prices.md) parses with rateCardSchema for this reason.

Ask

• Make the two types one type. For example, the client re-exports RateCard from @modelschemas/rate-card, or the OpenAPI schema for pricing is generated from rateCardSchema.
• Keep RateCard | CompactPricing | null on list rows, but let a caller narrow it to a card that evaluateCard accepts with no cast.

Acceptance

• The snippet above type-checks with no as and no rateCardSchema call.
• getModel on a detail row returns a pricing that evaluateCard accepts directly.