{"slug": "groq-route-intermittently-returns-json-validate-failed-with-strict-structured", "title": "Groq route intermittently returns json_validate_failed with strict structured outputs", "summary": "Hugging Face's Inference Providers route for openai/gpt-oss-20b:groq intermittently returns HTTP 400 errors with json_validate_failed and structured_generation_failed during strict structured outputs, with 0/5 and 1/5 success rates on two entity-review fixtures. Replacing the schema with singleton enums improved success to 3/3 and 2/3, but one failure persisted, indicating enum selection contributes but does not fully explain the issue. The user requests confirmation of the error origin and known reliability limitations for the route.", "body_md": "Hi Hugging Face team,\n\nWe are seeing reproducible HTTP 400 failures through Hugging Face Inference Providers on the exact route:\n\n`openai/gpt-oss-20b:groq`\n\nThe workload uses strict structured outputs for a deterministic entity-matching classification. The inputs are public and non-personal. We ran bounded diagnostics without retries, prompt tuning, or fixture changes.\n\nEvery failed call returned:\n\n`invalid_request_error`\n\n`json_validate_failed`\n\n`structured_generation_failed`\n\nFive repetitions per conservative `review`\n\nfixture:\n\n`entity-review-01`\n\n: 0/5 successful`entity-review-02`\n\n: 1/5 successfulThe one successful response matched the expected decision and confidence exactly. The same 188-byte strict output schema succeeded on a comparator. Request sizes were similar, so schema shape and request size do not explain the fixture-specific failures.\n\nFailed request IDs for `entity-review-01`\n\n:\n\n`req_01m0jacw8zemabfp437wn7hdt8`\n\n`req_01m0jacwxjeeva9k8rd84gc3pj`\n\n`req_01m0jacxdne98r2akb1janc4w8`\n\n`req_01m0jacy5qep39padbzhy7tzsp`\n\n`req_01m0jacyw4eqvrhwd1p4b0pfak`\n\nFailed request IDs for `entity-review-02`\n\n:\n\n`req_01m0jad0rvedvva48h0mg0h4qk`\n\n`req_01m0jad1dse1zsj83nvcjb40yd`\n\n`req_01m0jad2fcexc89hmwe2p34arb`\n\n`req_01m0jad329en9s172f1vrcqrvg`\n\nWe then kept the prompt and inputs unchanged but replaced the response schema in memory with singleton enums requiring the already-checked output: `match=review`\n\n, `confidence=low`\n\n.\n\nResults across three repetitions per fixture:\n\n`entity-review-01`\n\n: 3/3 successful`entity-review-02`\n\n: 2/3 successfulThe remaining failure had the same fingerprint even though the model had no enum choice:\n\n`req_01m0jzsav8e71rdt2j9qc4k27m`\n\nThis suggests enum selection contributes materially, but does not fully explain the route-level structured-generation failures.\n\nCould you please confirm:\n\n`json_validate_failed`\n\noriginated (HF router validation, provider validation, or generation)?`openai/gpt-oss-20b:groq`\n\nhas a known reliability limitation with strict JSON-schema outputs?The evidence is stored in a private repository, so the complete reproducible record is included here instead of linking to inaccessible CI pages.\n\nNormal-schema targeted diagnostic:\n\n`95c970d8f66979413bed18b680b86cfeeeec8780`\n\n`32490428886`\n\n`hf-benchmark-report-32490428886`\n\n(artifact `9449532403`\n\n)`sha256:c04babef1bb956d0b6f3633887521fbe6583d1e805bd85407dd998250ba1d7ef`\n\n`1.3.0`\n\nForced-singleton-schema diagnostic:\n\n`49ea63697f5afb82cccad93294ed858770c93232`\n\n`32523118374`\n\n`hf-benchmark-report-32523118374`\n\n(artifact `9461257213`\n\n)`sha256:da529cd0ef0fd601b9cae7dc948b72acdc6d3ce1ccf5be3f3ce1ffc0a486a769`\n\n`1.4.0`\n\nBoth diagnostics used the same policy (`phase-12-2-hugging-face-public-evaluation-v1`\n\n), registry (`hugging-face-structured-evaluation@1.1.0`\n\n), and fixtures (`public-business-controlled-ai-v1@1.0.0`\n\n). Raw prompts, inputs, outputs, provider messages, failed generations, and credentials were not retained. Authority leakage was zero.", "url": "https://wpnews.pro/news/groq-route-intermittently-returns-json-validate-failed-with-strict-structured", "canonical_source": "https://discuss.huggingface.co/t/groq-route-intermittently-returns-json-validate-failed-with-strict-structured-outputs/179097#post_1", "published_at": "2026-08-21 20:36:52+00:00", "updated_at": "2026-08-21 20:43:42.707872+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-products", "ai-infrastructure"], "entities": ["Hugging Face", "Groq", "openai/gpt-oss-20b:groq", "entity-review-01", "entity-review-02"], "alternates": {"html": "https://wpnews.pro/news/groq-route-intermittently-returns-json-validate-failed-with-strict-structured", "markdown": "https://wpnews.pro/news/groq-route-intermittently-returns-json-validate-failed-with-strict-structured.md", "text": "https://wpnews.pro/news/groq-route-intermittently-returns-json-validate-failed-with-strict-structured.txt", "jsonld": "https://wpnews.pro/news/groq-route-intermittently-returns-json-validate-failed-with-strict-structured.jsonld"}}