✓ PASS

vercel-ai 7.0.153

stream.error-midstream · gemini

Requirements

Id	Level	Status	Evidence
error-reported	MUST	SATISFIED	outcome=returned; error.category="api"; finish.value="error" (primary: fullStream.finish.finishReason); truth.finish="error"

What the provider sent

Finish

error

Raw finish

null

Text

Octop

Tool calls

[]

Usage

{
  "alternatives": {},
  "cached_input": null,
  "input": 22,
  "output": 2,
  "raw": {
    "candidatesTokenCount": 2,
    "promptTokenCount": 22,
    "promptTokensDetails": [
      {
        "modality": "TEXT",
        "tokenCount": 22
      }
    ],
    "serviceTier": "standard",
    "totalTokenCount": 24
  },
  "reasoning": null,
  "total": 24
}

Stream completed

false

Error

Message: This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.
Status: null
Type: UNAVAILABLE
Code: 503

Wire excerpt

status: 200
content-type: text/event-stream
body (sse):
data: {"candidates": [{"content": {"parts": [{"text": "Octop"}],"role": "model"},"index": 0}],"usageMetadata": {"promptTokenCount": 22,"candidatesTokenCount": 2,"totalTokenCount": 24,"promptTokensDetails": [{"modality": "TEXT","tokenCount": 22}],"serviceTier": "standard"},"modelVersion": "gemini-3.5-flash-lite","responseId": "pgTOaonbKMuV_uMPvK7d4AE"}


data: {"error":{"code":503,"message":"This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.","status":"UNAVAILABLE"}}

What the library reported

Outcome

returned

Finish

{
  "raw": null,
  "source": "fullStream.finish.finishReason",
  "surface": "primary",
  "value": "error"
}

Text

Value: Octop
Surface: primary

Tool calls

[]

Usage

{
  "cached_input": 0,
  "input": 22,
  "output": 2,
  "reasoning": 0,
  "surface": {
    "cached_input": "primary",
    "input": "primary",
    "output": "primary",
    "reasoning": "primary",
    "total": "primary"
  },
  "total": 24
}

Stream

{
  "chunks_seen": 8,
  "completed": false,
  "used": true
}

Error

Type: AI_StreamProviderError
Message: This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.
Category: api
Retry after: null
Stack top: ai/dist/index.js:8204

Unsupported reason

null

Full observation JSON

{
  "elapsed_ms": 133.52532499999998,
  "error": {
    "category": "api",
    "message": "This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.",
    "retry_after": null,
    "stack_top": "ai/dist/index.js:8204",
    "type": "AI_StreamProviderError"
  },
  "finish": {
    "raw": null,
    "source": "fullStream.finish.finishReason",
    "surface": "primary",
    "value": "error"
  },
  "format": "wiretruth.observation/v1",
  "library": {
    "api": "streamText",
    "language": "javascript",
    "name": "vercel-ai",
    "version": "7.0.153"
  },
  "notes": [],
  "outcome": "returned",
  "stream": {
    "chunks_seen": 8,
    "completed": false,
    "used": true
  },
  "structured": {
    "attempted": false,
    "returned_as_success": false,
    "value": null
  },
  "text": {
    "surface": "primary",
    "value": "Octop"
  },
  "tool_calls": [],
  "unsupported_reason": null,
  "usage": {
    "cached_input": 0,
    "input": 22,
    "output": 2,
    "reasoning": 0,
    "surface": {
      "cached_input": "primary",
      "input": "primary",
      "output": "primary",
      "reasoning": "primary",
      "total": "primary"
    },
    "total": 24
  }
}

Fixture

derived gemini/stream.error-midstream

Provenance	Value
Transform	inject_error_event
Source fixture	gemini/truncation.text-stream
Rationale	The provider reports an error after starting a successful HTTP stream.
Evidence	https://github.com/googleapis/python-genai/blob/main/google/genai/_api_client.py

Reproduce

wiretruth serve gemini/stream.error-midstream

Plain-library snippet from the shim manifest.

createGoogleGenerativeAI({ baseURL: WIRETRUTH_BASE_URL_GEMINI + '/v1beta', apiKey })(model)

Upstream

released Issue · PR

Fixed in: @ai-sdk/google 4.0.99

@ai-sdk/google's stream chunk schema has no error field, so an {"error": ...} SSE event after some content parses as an empty chunk: no error part, and the stream finishes with 'other'. Google's python-genai raises on such a chunk. Fixed by #23459 in @ai-sdk/google 4.0.99; the derived Gemini fixture passes on ai 7.0.146 (no live in-stream Gemini error has been observed).

History

Version	Grade
7.0.153	✓ PASS
7.0.146	✓ PASS
7.0.142	✗ WRONG