Guaranteed JSON, every single call

No dark-mode styles — this email renders the same in a dark client.

( https://app.requesty.ai/analytics )

Get guaranteed JSON, not a hope and a regex.
--------------------------------------------

Hi there
­

If you're parsing model responses with regex or hoping the JSON comes back well-formed, there's a better way. *Structured outputs* force the model to return JSON matching the exact fields and types you define, so downstream code can trust the shape every time.

Here's the part that only works through a gateway: you send *one OpenAI-style request* , and Requesty translates response_format into whatever each provider expects, Anthropic, Google, and the rest. Switch models or fail over, and your schema stays identical.

Add it with one parameter on the call you're already making:

response_format={
"type": "json_schema",
"json_schema": {
"name": "user_info",
"strict": True,
"schema": {
"type": "object",
"properties": {"name": {"type": "string"}, "email": {"type": "string"}},
"required": ["name", "email"],
"additionalProperties": False
}
}
}

We tested this across 244 models to make sure it holds up. It's best supported on OpenAI and Anthropic; a few providers only offer simpler JSON-object mode, so keep light error handling around parsing.

See the guide ( https://docs.requesty.ai/features/structured-outputs )

Full schema options and per-provider notes are in the structured outputs guide ( https://docs.requesty.ai/features/structured-outputs ). Questions? Just reply, or reach us at thibault@requesty.ai ( thibault@requesty.ai ).

Best regards,

Thibault Jaigu

Co-founder, Requesty

https://x.com/ThibaultJaigu ( https://x.com/ThibaultJaigu )

Empowering developers, one request at a time.

( https://www.linkedin.com/company/requesty ) ­

( https://www.youtube.com/@requestyAI ) ­

( https://discord.com/invite/Td3rwAHgt4 ) ­

( https://www.reddit.com/r/Requesty/ ) ­

( https://x.com/RequestyAI )

Unsubscribe ( https://jdjihac.r.bh.d.sendibt3.com/tr/un/v2/keX8ukzRdZozCFjRkxijVnNIf-fJajiDwBPOHlQdnUPEcczMHwGgyfdlVLx9uxf4Dyv3d5XrGPePUQjtBS-CMLDAzLRve5vAEdc1sBZEeKWeBUlPCjffJ24JX_Nz7zfMpP-cUF-3lH8uCEZliwWeoqCBDQJlgpbS7xjrAGOLLJISznQGzrM8AzQt29QgQTOnVg3n07BVd7onk44Z_JMTmESYs5WbgNTwoedCk9hjQKR5-5aWxO5CzAvvSS5HtrEvvTedjQN9knlYNh8Z51pYOgJjysqC26edLwQ-gY-Xq4InF0L5w1iHc0k1pislRPSd3O0YOxMbIGYTe0xWI3YapvqlBbUR5QVrTK4 )

Accession

nse/requesty/2026-11-02/5892

From

Requesty · thibault@requesty.ai

Sent

2 November 2026 · 10:00 UTC