JSON schema output fails under parallel load (100 requests)

Forum
1

Hi Groq team,

When I run 100 parallel requests to /openai/v1/chat/completions with response_format: { type: "json_schema" }, many responses fail JSON generation (invalid/non-JSON content / schema not respected). The same prompt works reliably when run sequentially or with low concurrency.

Is there a known issue or best-practice for strict JSON mode under high concurrency (limits, retries, different params, etc.)?

Thanks!

Résumé

Ce texte sera masqué

2

hello you can find everything to reprodude here GitHub - jsotogaviard/groq · GitHub

when i run node parallel i get lots of (specially after 3pm Paris time)

{ “request_id”: “req_01kjm4f7rffhdvaqqgez14s1k5”, “created_at”: “2026-03-01T07:23:12.015Z”, “error”: { “message”: “Generated JSON does not match the expected schema. Please adjust your prompt. See ‘failed_generation’ for more details. Error: jsonschema: ‘’ does not validate with /additionalProperties: additionalProperties ‘additionalProperties’, ‘properties’, ‘required’, ‘type’, ‘$schema’ not allowed”, “type”: “invalid_request_error”, “param”: “”, “code”: “json_validate_failed” } }

thanks for your help