{
  "checkedAt": "2026-08-15T05:27:51.561Z",
  "status": "output-budget-contract-verified",
  "scope": {
    "kind": "official-document extraction plus ten local token-budget fixtures",
    "authenticatedApiCalls": 0,
    "liveModelCalls": 0,
    "graderCalls": 0,
    "credentialsUsed": false,
    "warning": "Fixtures validate published limits and arithmetic only. They are not provider responses and do not prove route admission, billing, latency, tokenization parity, or output quality."
  },
  "sources": {
    "coreParameters": {
      "url": "https://docs.z.ai/guides/overview/concept-param.md",
      "status": 200,
      "bytes": 10543,
      "sha256": "b07051b257be57a382efdb536a72c12ddff7d332ac0776bb92db849da0423143",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    },
    "modelGuide": {
      "url": "https://docs.z.ai/guides/llm/glm-5.2.md",
      "status": 200,
      "bytes": 25293,
      "sha256": "eeca971119068b0619137ce9c9d3c9362131baf16fdab0c067eefe5802919023",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    },
    "chatCompletion": {
      "url": "https://docs.z.ai/api-reference/llm/chat-completion.md",
      "status": 200,
      "bytes": 42731,
      "sha256": "322c87b850c4504873b8c225b5b31266d90aa129f362134fa040c49c5943a7c9",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    },
    "modelApi": {
      "url": "https://huggingface.co/api/models/zai-org/GLM-5.2",
      "status": 200,
      "bytes": 23604,
      "sha256": "38fbba9fd2382e61c416d0a98a11d469629dc09ecdb468e20ecf3e0fd6ec1a24",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    },
    "zcodeIssue": {
      "url": "https://api.github.com/repos/zai-org/feedback/issues/96",
      "status": 200,
      "bytes": 8116,
      "sha256": "a24cfe94887e2314993d97b20f4b066c8a394f73546aa3f54a5ec63fa2e7c429",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    },
    "vllmRecipe": {
      "url": "https://api.github.com/repos/vllm-project/recipes/contents/models/zai-org/GLM-5.2.yaml",
      "status": 200,
      "bytes": 21904,
      "sha256": "5f5f95cf6acd5dfface1f01105187eafb4250343052f51c7f863e6c0508fced9",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    },
    "glm52Blog": {
      "url": "https://z.ai/blog/glm-5.2",
      "status": 200,
      "bytes": 598,
      "sha256": "63e94b9e11a2db64243ff18418011577f2a98ca7851ce9dcb0e97cfe34cc245a",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    },
    "zhipuResearch": {
      "url": "https://www.zhipuai.cn/zh/research",
      "status": 200,
      "bytes": 1243011,
      "sha256": "c9825ad32a505b312954e96db523aa18233301902cddb0a9b91244d939509402",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    },
    "pinnedConfig": {
      "url": "https://huggingface.co/zai-org/GLM-5.2/resolve/b4734de4facf877f85769a911abafc5283eab3d9/config.json",
      "status": 200,
      "bytes": 3732,
      "sha256": "185f93ee6d12548e16a847e279dc0c3c90b1524c970b0866b42fb545747d859a",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    },
    "pinnedReadme": {
      "url": "https://huggingface.co/zai-org/GLM-5.2/resolve/b4734de4facf877f85769a911abafc5283eab3d9/README.md",
      "status": 200,
      "bytes": 10905,
      "sha256": "ed5aca8ce3dc5f8de626c87e488444343e43b1dcbdeb0e643dc72fea63ab06e8",
      "transport": "curl with redirects, compression, a 30-second bound, and an identifiable user agent"
    }
  },
  "constants": {
    "nativeContextTokens": 1048576,
    "hostedDefaultOutputTokens": 65536,
    "hostedMaximumOutputTokens": 131072
  },
  "findings": {
    "hostedDefaultOutputTokens": 65536,
    "hostedMaximumOutputTokens": 131072,
    "modelGuideContextLabel": "1M",
    "modelGuideMaximumOutputLabel": "128K",
    "nativeMaxPositionEmbeddings": 1048576,
    "modelRevision": "b4734de4facf877f85769a911abafc5283eab3d9",
    "modelConfigTransformersVersion": "5.12.0",
    "coreParametersSayMaxTokensExcludesInput": true,
    "coreParametersDefineContextAsInputPlusGenerated": true,
    "chatCompletionMaximum": 131072,
    "completionUsageFields": {
      "promptTokens": true,
      "completionTokens": true,
      "totalTokens": true
    },
    "finishReasons": [
      "stop",
      "tool_calls",
      "length",
      "sensitive",
      "model_context_window_exceeded",
      "network_error"
    ],
    "issue": {
      "repository": "zai-org/feedback",
      "number": 96,
      "state": "open",
      "createdAt": "2026-07-07T00:24:15Z",
      "authorAssociation": "NONE",
      "isUserSubmittedReport": true,
      "reportsContextValuesMappedToMaxTokens": true,
      "reports200kRejected": true,
      "reports1mRejected": true,
      "reports131072Accepted": true
    },
    "vllmRecipe": {
      "contentSha": "c0d146d82ff822486ad0760d07fd6954fab3c18d",
      "minVllmVersion": "0.23.0",
      "nativeContextTokens": 1048576,
      "standardExampleServedContextTokens": 131072,
      "exampleRequestMaxTokens": 4096
    },
    "fixedDiscoveryChecks": {
      "glm52BlogReturned200": true,
      "glm52BlogNamesGlm52": false,
      "zhipuResearchReturned200": true,
      "zhipuResearchNamesGlm52": true
    }
  },
  "fixtures": {
    "hostedDefault": {
      "accepted": true,
      "errors": [],
      "inputTokens": 100000,
      "requestedOutputTokens": 65536,
      "reserveTokens": 0,
      "plannedTotalTokens": 165536,
      "effectiveContextTokens": 1048576,
      "effectiveOutputLimitTokens": 131072,
      "remainingContextTokens": 948576
    },
    "exactOutputCeiling": {
      "accepted": true,
      "errors": [],
      "inputTokens": 400000,
      "requestedOutputTokens": 131072,
      "reserveTokens": 0,
      "plannedTotalTokens": 531072,
      "effectiveContextTokens": 1048576,
      "effectiveOutputLimitTokens": 131072,
      "remainingContextTokens": 648576
    },
    "exactNativeBoundary": {
      "accepted": true,
      "errors": [],
      "inputTokens": 917504,
      "requestedOutputTokens": 131072,
      "reserveTokens": 0,
      "plannedTotalTokens": 1048576,
      "effectiveContextTokens": 1048576,
      "effectiveOutputLimitTokens": 131072,
      "remainingContextTokens": 131072
    },
    "oneAboveOutputCeiling": {
      "accepted": false,
      "errors": [
        "OUTPUT_LIMIT_EXCEEDED"
      ],
      "inputTokens": 1000,
      "requestedOutputTokens": 131073,
      "reserveTokens": 0,
      "plannedTotalTokens": 132073,
      "effectiveContextTokens": 1048576,
      "effectiveOutputLimitTokens": 131072,
      "remainingContextTokens": 1047576
    },
    "contextMappedToMax200k": {
      "accepted": false,
      "errors": [
        "OUTPUT_LIMIT_EXCEEDED"
      ],
      "inputTokens": 1000,
      "requestedOutputTokens": 200000,
      "reserveTokens": 0,
      "plannedTotalTokens": 201000,
      "effectiveContextTokens": 1048576,
      "effectiveOutputLimitTokens": 131072,
      "remainingContextTokens": 1047576
    },
    "contextMappedToMax1m": {
      "accepted": false,
      "errors": [
        "OUTPUT_LIMIT_EXCEEDED"
      ],
      "inputTokens": 0,
      "requestedOutputTokens": 1000000,
      "reserveTokens": 0,
      "plannedTotalTokens": 1000000,
      "effectiveContextTokens": 1048576,
      "effectiveOutputLimitTokens": 131072,
      "remainingContextTokens": 1048576
    },
    "defaultNearFullContext": {
      "accepted": false,
      "errors": [
        "CONTEXT_BUDGET_EXCEEDED"
      ],
      "inputTokens": 990000,
      "requestedOutputTokens": 65536,
      "reserveTokens": 0,
      "plannedTotalTokens": 1055536,
      "effectiveContextTokens": 1048576,
      "effectiveOutputLimitTokens": 131072,
      "remainingContextTokens": 58576
    },
    "lowerServedContext": {
      "accepted": false,
      "errors": [
        "CONTEXT_BUDGET_EXCEEDED"
      ],
      "inputTokens": 150000,
      "requestedOutputTokens": 64000,
      "reserveTokens": 0,
      "plannedTotalTokens": 214000,
      "effectiveContextTokens": 200000,
      "effectiveOutputLimitTokens": 131072,
      "remainingContextTokens": 50000
    },
    "lowerClientOutputCap": {
      "accepted": false,
      "errors": [
        "OUTPUT_LIMIT_EXCEEDED"
      ],
      "inputTokens": 10000,
      "requestedOutputTokens": 128000,
      "reserveTokens": 0,
      "plannedTotalTokens": 138000,
      "effectiveContextTokens": 1048576,
      "effectiveOutputLimitTokens": 64000,
      "remainingContextTokens": 1038576
    },
    "reservedContextOverflow": {
      "accepted": false,
      "errors": [
        "CONTEXT_BUDGET_EXCEEDED"
      ],
      "inputTokens": 950000,
      "requestedOutputTokens": 90000,
      "reserveTokens": 10000,
      "plannedTotalTokens": 1050000,
      "effectiveContextTokens": 1048576,
      "effectiveOutputLimitTokens": 131072,
      "remainingContextTokens": 88576
    }
  },
  "invariants": {
    "requiredSourcesReturned200": true,
    "pinnedModelFilesReturned200": true,
    "documentedDefaultMatchesFixture": true,
    "documentedMaximumMatchesFixture": true,
    "nativeContextMatchesFixture": true,
    "chatContractMatchesDocumentedMaximum": true,
    "exactBoundariesPass": true,
    "outputOverflowFixturesFailOutputGate": true,
    "contextOverflowFixturesFailContextGate": true,
    "lowerClientCapFailsOutputGate": true,
    "issueDescribesTheSameMappingFailure": true
  },
  "decisionRule": {
    "effectiveContextTokens": "min(native context, route or server context, client context)",
    "effectiveOutputLimitTokens": "min(model output ceiling, route output limit, client output limit)",
    "acceptedWhen": [
      "1 <= requested output <= effective output limit",
      "input + requested output + reserve <= effective context"
    ]
  },
  "limitations": [
    "The Z.AI default and maximum are documentation values observed on the check date; route-specific limits can be lower.",
    "The 1,048,576-token config field is a native checkpoint property, not proof that every hosted route exposes the full window.",
    "The ZCode issue is a user-submitted report in an official repository, not a provider-authenticated response captured by this audit.",
    "No tokenizer call was made. Production should measure the exact rendered prompt and reserve space for templates, tools, and retries.",
    "The public Zhipu research index is a required discovery check; a bounded timeout is recorded rather than treated as corroborating evidence if unavailable."
  ]
}