Inference Providers

An inference provider serves model requests. It can serve models trained by many developers. The public directory includes reviewed platforms with indexed offerings, ordered by indexed model count. Zero-offering providers are excluded from the directory.

The Web model library, organization directories and their model catalogs display 18 entries per page. API consumers can request their own page size with limit and use offset to paginate.

Use the canonical catalog endpoint for fused model cards. The source-native models endpoint retains the provider’s original identifiers and metadata, including deployment versions. These two representations answer different questions and must not be counted as interchangeable models.

Cheaper Inference (cheaperInference) is an inference gateway, separate from the model creator. Its local collector uses the documented, keyless https://api.cheaperinference.com/public/models catalog on a 15-minute cadence. The native cached response keeps the upstream models wrapper. This public catalog does not establish coverage of authenticated marketplace-only models. Metadata is retained across supported modalities; the initial price adapter publishes only validated, single-tier standard text offerings. Ambiguous tiers, unknown units and unavailable entries retain metadata without comparable prices.

Service status comes from official structured status feeds. It describes reported service/component state and observation freshness; it is not a measured per-model uptime guarantee. Status history is a seven-day record of checks and reports. Original status evidence is retrieved by its recorded checksum.

Collected organization introductions are available as nullable profile objects. They include summary, paragraph-separated plain-text description, homepage, links, source_url, source_kind, retrieved_at, receipt_id and parser_version. The existing top-level description is the short introduction or an empty string. Omit the introduction when profile is null. Collection happens in the serial worker, never while serving an API request.

On this page

Read a source-native provider catalog

GET /api/v1/providers/{provider}/models

Server-cached provider metadata. Model Garden, Amazon Bedrock, Microsoft Foundry, NVIDIA, Black Forest Labs, Meta, ARC, Supernova, Alibaba Cloud Model Studio and Cheaper Inference expose pagination with total, has_more, next_offset and receipt_id. Pin receipt_id from the first page for an immutable snapshot across refreshes. Foundry includes all public nonarchived registry/model versions, including third-party and Hub models; Bedrock foundation models are scoped to the configured AWS region. Supernova offerings preserve standard/priority API IDs, base-model associations and public effective USD quotes; paused offerings retain metadata without purchasable prices. Alibaba Cloud is scoped to Singapore / international; publisher, inference provider, snapshot aliases, tiered catalog quotes and unverified currencies remain explicit. Cheaper Inference exposes its keyless public catalog; gateway attribution stays separate from model creators, and only validated single-tier standard text quotes are comparable. Credentials stay on the server. Deployment examples and hardware do not establish model limits or unit prices.

Access: public read; no TensorTrace API key.

Parameters

NameLocationRequiredType / limitsMeaning
limitqueryNointeger Default: 500; Minimum: 1; Maximum: 1000Maximum number of returned rows. The documented default and range apply; numeric pagination limits are clamped unless stated otherwise.
offsetqueryNointeger Default: 0; Minimum: 0Zero-based number of matching rows to skip. Default 0; negative values are clamped to 0.
providerpathYesstringRegistered inference-provider/source ID, such as openRouter.
receipt_idqueryNostring · uuidUUID of an immutable collection receipt. Omit it to use the latest source receipt.

Request example

curl 'https://tensortrace.net/api/v1/providers/openai/models?limit=1'

Responses

StatusMeaningBody type
200Source-backed JSON responseNativeProviderModelsResponse
400Invalid queryNo declared JSON body
404Unknown model/sourceNo declared JSON body
500Database unavailableNo declared JSON body
503Service temporarily unavailableNo declared JSON body

Response example · 200

Illustrative values; schema-valid and not a live catalog measurement.

{}

Type: NativeProviderModelsResponse. Provider-compatible cached catalog. DeepInfra uses a top-level array; other providers retain their native object wrappers. Each cached model object includes confirmed TensorTrace identity metadata.

Current official status for all reviewed serving platforms

GET /api/v1/providers/status

Access: public read; no TensorTrace API key.

Parameters

This operation has no path or query parameters.

Request example

curl 'https://tensortrace.net/api/v1/providers/status'

Responses

StatusMeaningBody type
200Official provider status responseobject
400Invalid queryNo declared JSON body
404Unknown provider or expired evidenceNo declared JSON body

Response example · 200

Illustrative values; schema-valid and not a live catalog measurement.

{
    "data": [
        {
            "source_id": "openRouter",
            "display_name": "Model Reviewer",
            "state": "operational",
            "freshness": "fresh",
            "scope": "example",
            "feed_urls": [
                "https://example.com/model"
            ],
            "poll_interval_seconds": 1,
            "scheduler_enabled": false,
            "history_retention_days": 1
        }
    ],
    "as_of": "2026-10-05T00:00:00Z",
    "measurement": "example",
    "model_endpoint_health": false
}
FieldTypeRequiredMeaning / constraints
dataarray of ProviderStatusYes
as_ofstring · date-timeYes
measurementstringYes
model_endpoint_healthbooleanYes

Current scoped provider status

GET /api/v1/providers/{provider}/status

Access: public read; no TensorTrace API key.

Parameters

NameLocationRequiredType / limitsMeaning
providerpathYesstringRegistered inference-provider/source ID, such as openRouter.

Request example

curl 'https://tensortrace.net/api/v1/providers/openai/status'

Responses

StatusMeaningBody type
200Official provider status responseProviderStatus
400Invalid queryNo declared JSON body
404Unknown provider or expired evidenceNo declared JSON body

Response example · 200

Illustrative values; schema-valid and not a live catalog measurement.

{
    "source_id": "openRouter",
    "display_name": "Model Reviewer",
    "state": "operational",
    "freshness": "fresh",
    "scope": "example",
    "feed_urls": [
        "https://example.com/model"
    ],
    "poll_interval_seconds": 1,
    "scheduler_enabled": false,
    "history_retention_days": 1
}
FieldTypeRequiredMeaning / constraints
source_idstringYes
display_namestringYes
scopestringYes
page_urlstring,null · uriNo
formatstring,nullNo
feed_urlsarray of string · uriYes
component_idsarray of stringNo
group_idsarray of stringNo
notestring,nullNo
statestringYesValues: operational, degraded, partial_outage, major_outage, maintenance, incident_reported, no_reported_incidents, unknown
freshnessstringYesValues: fresh, stale, fetch_failed, not_checked, unverified
reported_statestring,nullNo
checked_atstring,null · date-timeNo
last_success_atstring,null · date-timeNo
expires_atstring,null · date-timeNo
consecutive_failuresintegerNo
error_messagestring,nullNo
snapshotobject,nullNo
evidencearray of objectNo
poll_interval_secondsintegerYes
scheduler_enabledbooleanYes
history_retention_daysintegerYes

Seven-day check history; not a measured uptime percentage

GET /api/v1/providers/{provider}/status/history

Access: public read; no TensorTrace API key.

Parameters

NameLocationRequiredType / limitsMeaning
providerpathYesstringRegistered inference-provider/source ID, such as openRouter.
limitqueryNointeger Default: 20; Minimum: 1; Maximum: 100Number of check observations to return. Default 20; values outside 1–100 return 400.
beforequeryNointeger Minimum: 1Exclusive observation ID cursor returned by the status history. Must be a positive integer.

Request example

curl 'https://tensortrace.net/api/v1/providers/openai/status/history?limit=1'

Responses

StatusMeaningBody type
200Official provider status responseobject
400Invalid queryNo declared JSON body
404Unknown provider or expired evidenceNo declared JSON body

Response example · 200

Illustrative values; schema-valid and not a live catalog measurement.

{
    "data": [
        {}
    ],
    "next_before": 1,
    "retention_days": 1
}
FieldTypeRequiredMeaning / constraints
dataarray of objectNo
next_beforeinteger,nullNo
retention_daysintegerNo

Original upstream JSON bytes, SHA-256 verified; AWS uses BOM-declared UTF-16

GET /api/v1/providers/{provider}/status/evidence/{checksum}

Access: public read; no TensorTrace API key.

Parameters

NameLocationRequiredType / limitsMeaning
providerpathYesstringRegistered inference-provider/source ID, such as openRouter.
checksumpathYesstring Pattern: ^[a-fA-F0-9]{64}$Registered inference-provider/source ID, such as openRouter.

Request example

curl 'https://tensortrace.net/api/v1/providers/openai/status/evidence/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa'

Responses

StatusMeaningBody type
200Official provider status responseany JSON
400Invalid queryNo declared JSON body
404Unknown provider or expired evidenceNo declared JSON body

Response example · 200

Illustrative values; schema-valid and not a live catalog measurement.

"example"

Type: any JSON.

List inference providers

GET /api/v1/providers

Access: public read; no TensorTrace API key.

Parameters

NameLocationRequiredType / limitsMeaning
qqueryNostringCase-insensitive search text. Omit it to browse the complete matching catalog.
limitqueryNointeger Default: 48; Minimum: 1; Maximum: 100Maximum number of returned rows. The documented default and range apply; numeric pagination limits are clamped unless stated otherwise.
offsetqueryNointeger Default: 0; Minimum: 0Zero-based number of matching rows to skip. Default 0; negative values are clamped to 0.

Request example

curl 'https://tensortrace.net/api/v1/providers?limit=1'

Responses

StatusMeaningBody type
200Incrementally prepared local dataDirectoryPage
400Invalid queryErrorResponse
404Not present in the prepared catalogErrorResponse
503Catalog preparation has not completed yetErrorResponse

Response example · 200

Illustrative values; schema-valid and not a live catalog measurement.

{
    "data": [
        {
            "id": "openai",
            "name": "OpenAI",
            "role": "inference_provider",
            "model_count": 1,
            "description": "Illustrative model observation; consult live evidence for current specifications.",
            "catalog_scope": "example"
        }
    ],
    "total_count": 1,
    "limit": 1,
    "offset": 0
}
FieldTypeRequiredMeaning / constraints
dataarray of OrganizationYesMax. items: 100
total_countintegerYesMinimum: 0
limitintegerYesMinimum: 1; Maximum: 100
offsetintegerYesMinimum: 0

Read an inference provider

GET /api/v1/providers/{provider}

Access: public read; no TensorTrace API key.

Parameters

NameLocationRequiredType / limitsMeaning
providerpathYesstringRegistered inference-provider/source ID, such as openRouter.

Request example

curl 'https://tensortrace.net/api/v1/providers/openai'

Responses

StatusMeaningBody type
200Incrementally prepared local dataOrganization
400Invalid queryErrorResponse
404Not present in the prepared catalogErrorResponse
503Catalog preparation has not completed yetErrorResponse

Response example · 200

Illustrative values; schema-valid and not a live catalog measurement.

{
    "id": "openai",
    "name": "OpenAI",
    "role": "inference_provider",
    "model_count": 1,
    "description": "Illustrative model observation; consult live evidence for current specifications.",
    "catalog_scope": "example"
}
FieldTypeRequiredMeaning / constraints
idstringYes
namestringYes
rolestringYesValues: developer, inference_provider
model_countintegerYesMinimum: 0
descriptionstringYesCollected short introduction; empty when no usable official description has been collected.
catalog_scopestringYes
aliasesarray of stringNo
logo_urlstringNo
homepagestring,nullNo
source_statusstringNo
observed_atstring,null · date-timeNo
offering_countintegerNo
priced_model_countintegerNo
best_modelobject,nullNo
governancearray of OrganizationGovernanceFindingNoDeveloper company disclosure assessments, prepared independently of model matches. Empty when no assessment is recorded; omitted for inference providers.
leading_modelsarray of ProviderLeadingModelNoUp to three leaders from the developer's models or inference provider's catalog, ordered by intelligence, capability, benchmark rating, name and canonical ID. Publisher aliases are combined for developers and each canonical model appears once. Empty when no models are available; omitted by older prepared directories. Developer best_model is retained as the first entry for compatibility. Max. items: 3
profileOrganizationProfile or nullNoWorker-collected company or inference-service profile. Null when unavailable; collection never runs in API requests.

List a provider’s canonical models

GET /api/v1/providers/{provider}/catalog

Access: public read; no TensorTrace API key.

Parameters

NameLocationRequiredType / limitsMeaning
providerpathYesstringRegistered inference-provider/source ID, such as openRouter.
qqueryNostringCase-insensitive search text. Omit it to browse the complete matching catalog.
queryqueryNostringAlternative spelling of q; when both are supplied, query takes precedence.
providerqueryNostringModel publisher namespace. For serving-platform filters, use inference_provider instead.
sourcequeryNostringRegistered data-source ID, such as openRouter. This identifies the collecting source, not necessarily the model developer.
pricequeryNostringPricing class: all, free or paid. Uses comparable basic token prices; an unknown price is not free.
capabilityqueryNostringComma-separated capability identifiers; all must match.
input_modalityqueryNostringMatch an input modality, such as text, image, audio or video. Obtain current values from /facets/models.
model_typequeryNostring Max. length: 64Match a displayed model classification, including secondary types. Use multimodal for MLLM, vlm for image-only VLM, llm for text LLM, or another model type ID from a classification link. unknown selects models without reported classification evidence. Omit or use all for no type filter. Combines with other discovery filters before pagination and counts; an unrecognized ID returns an empty result.
output_modalitiesqueryNostringComma-separated output modalities; matches models with any listed output modality. all disables this filter.
sortqueryNostring Default: "intelligence"Defaults to Intelligence Index descending; unscored models follow using capability, rating and name as deterministic fallbacks. Price sorts use the card-selected USD Input then Output rates, with missing rates last; cache rates do not participate. Tied basic offerings prefer the model developer. Provider-scoped sorts apply the same policy within that source's quotes.
limitqueryNointeger Default: 48; Minimum: 1; Maximum: 100Maximum number of returned rows. The documented default and range apply; numeric pagination limits are clamped unless stated otherwise.
offsetqueryNointeger Default: 0; Minimum: 0Zero-based number of matching rows to skip. Default 0; negative values are clamped to 0.
identity_kindqueryNostring Values: model, familyRestrict identities to a concrete model or a model family; omit to include both.
developerqueryNostringCanonical developer ID; groups reviewed publisher aliases.
inference_providerqueryNostringRegistered inference source ID; selects current source offerings, including price-only records. Adds source-scoped provider_prices; other model fields retain canonical scope.

Request example

curl 'https://tensortrace.net/api/v1/providers/openai/catalog?limit=1'

Responses

StatusMeaningBody type
200Incrementally prepared local dataModelPage
400Invalid queryErrorResponse
404Not present in the prepared catalogErrorResponse
503Catalog preparation has not completed yetErrorResponse

Response example · 200

Illustrative values; schema-valid and not a live catalog measurement.

{
    "data": [
        {
            "id": "openai/gpt-4.1",
            "canonical_model_id": "openai/gpt-4.1",
            "provider": "openai",
            "name": "OpenAI: GPT-4.1",
            "official_model_id": "gpt-4.1",
            "developer_id": "openai",
            "developer_name": "OpenAI",
            "description": "Illustrative model observation; consult live evidence for current specifications.",
            "context_length": 1048576
        }
    ],
    "total_count": 1,
    "limit": 1,
    "offset": 0
}
FieldTypeRequiredMeaning / constraints
dataarray of CatalogModelYesMax. items: 100
total_countintegerYesMinimum: 0
limitintegerYesMinimum: 1; Maximum: 100
offsetintegerYesMinimum: 0

Models

Loading
EVIDENCE / COMPARISON

Benchmarks

Explore every benchmark

Loading

These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

Install TensorTrace