✓ PASS

pydantic-ai 2.61.0

truncation.reasoning-only · gemini

Requirements

Id	Level	Status	Evidence
cause-is-truncation	MUST	SATISFIED	finish.value="length" (raw: ModelResponse.finish_reason); truth.finish="length"; error.category="truncation"; structured.returned_as_success=false
text-reported	MUST	SATISFIED	text.value=null (absent); truth.text=null
usage-reported	SHOULD	NOT_OBSERVABLE	usage.output=28 (raw); truth.usage.output=28

What the provider sent

Finish: length
Raw finish: MAX_TOKENS
Text: null
Tool calls: []
Usage: { "alternatives": {}, "cached_input": null, "input": 44, "output": 28, "raw": { "promptTokenCount": 44, "promptTokensDetails": [ { "modality": "TEXT", "tokenCount": 44 } ], "serviceTier": "standard", "thoughtsTokenCount": 28, "totalTokenCount": 72 }, "reasoning": 28, "total": 72 }
Stream completed: null
Error: null

Wire excerpt

status: 200
content-type: application/json; charset=UTF-8
body (json):
{
  "candidates": [
    {
      "content": {},
      "finishReason": "MAX_TOKENS",
      "index": 0
    }
  ],
  "usageMetadata": {
    "promptTokenCount": 44,
    "totalTokenCount": 72,
    "promptTokensDetails": [
      {
        "modality": "TEXT",
        "tokenCount": 44
      }
    ],
    "thoughtsTokenCount": 28,
    "serviceTier": "standard"
  },
  "modelVersion": "gemini-3.5-flash-lite",
  "responseId": "qAXOar_qAaeVmtkPp5_qyQQ"
}

What the library reported

Outcome

raised

Finish

{
  "raw": "MAX_TOKENS",
  "source": "ModelResponse.finish_reason",
  "surface": "raw",
  "value": "length"
}

Text

Value: null
Surface: absent

Tool calls

[]

Usage

{
  "cached_input": 0,
  "input": 44,
  "output": 28,
  "reasoning": 28,
  "surface": {
    "cached_input": "raw",
    "input": "raw",
    "output": "raw",
    "reasoning": "raw",
    "total": "raw"
  },
  "total": 72
}

Stream

{
  "chunks_seen": null,
  "completed": null,
  "used": false
}

Error

Type: pydantic_ai.exceptions.UnexpectedModelBehavior
Message: Model token limit (32) exceeded before any response was generated. Increase the `max_tokens` model setting, or simplify the prompt to result in a shorter response that will fit within the limit.
Category: truncation
Retry after: null
Stack top: pydantic_ai/_agent_graph.py:2495

Unsupported reason

null

Full observation JSON

{
  "elapsed_ms": 329.4894499704242,
  "error": {
    "category": "truncation",
    "message": "Model token limit (32) exceeded before any response was generated. Increase the `max_tokens` model setting, or simplify the prompt to result in a shorter response that will fit within the limit.",
    "retry_after": null,
    "stack_top": "pydantic_ai/_agent_graph.py:2495",
    "type": "pydantic_ai.exceptions.UnexpectedModelBehavior"
  },
  "finish": {
    "raw": "MAX_TOKENS",
    "source": "ModelResponse.finish_reason",
    "surface": "raw",
    "value": "length"
  },
  "format": "wiretruth.observation/v1",
  "library": {
    "api": "Agent.run",
    "language": "python",
    "name": "pydantic-ai",
    "version": "2.61.0"
  },
  "notes": [],
  "outcome": "raised",
  "stream": {
    "chunks_seen": null,
    "completed": null,
    "used": false
  },
  "structured": {
    "attempted": false,
    "returned_as_success": false,
    "value": null
  },
  "text": {
    "surface": "absent",
    "value": null
  },
  "tool_calls": [],
  "unsupported_reason": null,
  "usage": {
    "cached_input": 0,
    "input": 44,
    "output": 28,
    "reasoning": 28,
    "surface": {
      "cached_input": "raw",
      "input": "raw",
      "output": "raw",
      "reasoning": "raw",
      "total": "raw"
    },
    "total": 72
  }
}

Fixture

recorded gemini/truncation.reasoning-only

Provenance	Value
Recorded at	2026-10-13T10:19:19Z
Requested model	gemini-3.5-flash-lite
Returned model	gemini-3.5-flash-lite
CI run	https://github.com/kartsan03/wiretruth/actions/runs/38382529221
Notes	Recorded on gemini-3.5-flash-lite at thinkingLevel HIGH with maxOutputTokens 32, because gemini-3.8-flash answered HTTP 503 (high demand) on three attempts. Thoughts used 28 of the 32 tokens and the candidate has content {} with finishReason MAX_TOKENS: the output limit bounds thoughts and answer together.

Reproduce

wiretruth serve gemini/truncation.reasoning-only

Plain-library snippet from the shim manifest.

GoogleModel(model, provider=GoogleProvider(base_url=WIRETRUTH_BASE_URL_GEMINI, api_key=WIRETRUTH_API_KEY)); host root

Upstream

No matching upstream report.

History

Version	Grade
2.61.0	✓ PASS
2.56.0	✓ PASS