Infrastructure

typesafe-ai-benchmark

@iammrduncan32TypeScriptMITupdated 2026-09-19runnable

This is a LLM Gateway that mimics typesafe ai structured output. Like an imposter Jev.

iammrduncan/typesafe-ai-benchmark

Where it calls Jev

const isChat = (request: FastifyRequest) => request.url.split('?')[0] !== '/v1/systemone';

packages/api/src/http.ts:28

The link points at the commit we read, so the line number still holds.

What it asks Jev

These question sets are lifted from this project's source exactly as written.

from packages/api/examples/benchmark.ts:24

  1. refundtrue/false

    Does the customer request money back?

Run it on your own text

The code
import { experimental_evaluate as evaluate } from 'ai';

const { answers } = await evaluate({
  model: 'typesafe-ai/jev',
  state,
  questions: {
    refund: { type: 'boolean', instructions: 'Does the customer request money back?' },
  },
});

from packages/api/src/logging.test.ts:21

  1. CANARY_QUESTION_IDtrue/false

    CANARY_RUBRIC

Run it on your own text

The code
import { experimental_evaluate as evaluate } from 'ai';

const { answers } = await evaluate({
  model: 'typesafe-ai/jev',
  state,
  questions: {
    CANARY_QUESTION_ID: { type: 'boolean', instructions: 'CANARY_RUBRIC' },
  },
});

from packages/api/src/test-support.ts:8

  1. refundtrue/false

    Does the customer request a refund?

Run it on your own text

The code
import { experimental_evaluate as evaluate } from 'ai';

const { answers } = await evaluate({
  model: 'typesafe-ai/jev',
  state,
  questions: {
    refund: { type: 'boolean', instructions: 'Does the customer request a refund?' },
  },
});

Other projects in this category