AI API receipt · vllm-project/vllm
vllm-project/vllm#60998 · upstream bug reproduced by running the issue's own reproduction
✓ Badgr rejects this response
generate() with max_tokens=2.5 never returned: the engine logged msgspec.ValidationError and the client got no response and no error (reproduced on vLLM 0.30.0, RTX 4090; capped at 240 s).
Outcome contract verdict: stream ended without a terminal [DONE] or message_stop event. The attempt is recorded as failed (invalid_outcome) instead of a verified success, so it is not billed as a good answer and can fall back to another route.
The output the bug produces, taken from the issue’s reproduction and fed to Badgr’s validators. Not a live provider call.
(no events: the stream produced nothing)
Written by Badgr’s receipt ledger and read back from GET /v1/receipts/replay-vllm-60998-float-max-tokens-hangs.
{
"request_id": "replay-vllm-60998-float-max-tokens-hangs",
"status": "failed",
"endpoint": "/v1/completions",
"stream": true,
"failure_stage": "outcome_contract",
"error_code": "invalid_outcome",
"error_message": "Response did not satisfy the outcome contract: stream ended without a terminal [DONE] or message_stop event",
"retryable": true,
"provider_mode": "replay",
"routing_decision": "replayed_upstream_output",
"created_at": "1791679612.988222"
}Recorded 2026-10-11 in Badgr’s local development environment. Replay cases and verdicts are pinned by backend/test_ai_api_replay_cases.py.
← All AI API receipts