Prices
Prices belong to an offering on a serving platform. Complete model prices retain currency, billing unit, tiers, region, workload, cache reads/writes, batch rates, media units and source observations where reported. Use these fields to compare equivalent conditions.
The global price endpoint exposes source-backed USD quote observations. The model-specific endpoint pages through the full published quote set, with platform profiles. provider on that model-specific endpoint means a serving source ID, while the path’s publisher segment identifies the model.
Model cards select an offering by ordinary Input price, then Output price, with unknown values last. Equal basic prices prefer the model developer’s offering. Cached and conditional rates do not decide that selection. This does not claim that one provider is cheapest for every workload.
Monetary labels use two fixed decimal places with half-up rounding. Only when a
positive price would round to zero does display precision increase to four, six
or more even places, stopping at the first nonzero rounded result. For example,
1 displays as 1.00, 0.009 as 0.01, and 0.003 as 0.0030. This is a
presentation rule; the API, stored prices, comparisons and sort keys retain
their original precision.
Hovering a monetary label on the website reveals its full unrounded source/API
value, including the currency. Existing source and billing-condition hints are
retained alongside that value.
Cheaper Inference quotes retain the standard service tier, Chat Completions
protocol and zdr: false default scope. Standard data retention does not state
whether data is used for training. Scoped upstream rates are per token and are
converted once to the API's per-million units; flat catalog rates and different
context/service tiers are never mixed. Cache writes retain their 5-minute or
1-hour billing label. Source scope/version and effective/expiry times stay in
the original quote evidence. Expired quotes are retired during the serial
worker's catalog maintenance, while source metadata and receipts remain.
On this page
- Compare price observations ·
GET /prices - Page complete published inference prices for a canonical model ·
GET /models/{provider}/{model}/prices
Compare price observations
GET /api/v1/prices
Comparable USD amounts with explicit pricing units.
Access: public read; no TensorTrace API key.
Parameters
| Name | Location | Required | Type / limits | Meaning |
|---|---|---|---|---|
q | query | No | string | Case-insensitive search text. Omit it to browse the complete matching catalog. |
provider | query | No | string | Registered serving-source ID, such as anthropic or openRouter. Omit to include all platforms for this model. |
source | query | No | string | Registered data-source ID, such as openRouter. This identifies the collecting source, not necessarily the model developer. |
platform | query | No | string | Exact serving-platform label on a published price quote. |
unit | query | No | string | Billing or performance unit, preserving the source’s explicit denominator. Compare only equivalent units. |
sort | query | No | string Default: "amount-low-to-high"; Values: amount-low-to-high, amount-high-to-low, updated-newest, model, priceHigh, pricing-high-to-low, updated, name | Order of results. Missing comparable values sort after known values. |
limit | query | No | integer Default: 100; Minimum: 1; Maximum: 1000 | Maximum number of returned rows. The documented default and range apply; numeric pagination limits are clamped unless stated otherwise. |
offset | query | No | integer Default: 0; Minimum: 0 | Zero-based number of matching rows to skip. Default 0; negative values are clamped to 0. |
Request example
curl 'https://tensortrace.net/api/v1/prices?limit=1'
Responses
| Status | Meaning | Body type |
|---|---|---|
200 | Source-backed JSON response | PriceComparisonResponse |
400 | Invalid query | No declared JSON body |
404 | Unknown model/source | No declared JSON body |
500 | Database unavailable | No declared JSON body |
503 | Service temporarily unavailable | No declared JSON body |
Response example · 200
Illustrative values; schema-valid and not a live catalog measurement.
{
"data": [
{
"canonical_model_id": "google/gemini-2.5-flash",
"provider": "google",
"display_name": "Model Reviewer",
"source_id": "openRouter",
"platform": "OpenRouter",
"source_model_id": "google/gemini-2.5-flash",
"unit": "input_tokens_per_million",
"amount_usd": 0.3,
"currency": "USD",
"updated_at": "2026-10-05T00:00:00Z"
}
],
"total_count": 1
}
| Field | Type | Required | Meaning / constraints |
|---|---|---|---|
data | array of PriceComparisonRowDto | Yes | |
total_count | integer | Yes |
Page complete published inference prices for a canonical model
GET /api/v1/models/{provider}/{model}/prices
Preserves currency, unit, conditions, source receipt and observation time. Missing price data is an empty list, never a zero price. Query publications may change between requests; use publication_id to fence pagination.
Access: public read; no TensorTrace API key.
Parameters
| Name | Location | Required | Type / limits | Meaning |
|---|---|---|---|---|
provider | query | No | string | Registered serving-source ID, such as anthropic or openRouter. Omit to include all platforms for this model. |
limit | query | No | integer Default: 50; Maximum: 100 | Maximum number of returned rows. The documented default and range apply; numeric pagination limits are clamped unless stated otherwise. |
offset | query | No | integer Default: 0 | Zero-based number of matching rows to skip. Default 0; negative values are clamped to 0. |
provider | path | Yes | string | Canonical model publisher namespace; URL-encode this path segment. |
model | path | Yes | string | Canonical model name within the publisher namespace. URL-encode nested slashes in this single segment. |
Request example
curl 'https://tensortrace.net/api/v1/models/google/gemini-2.5-flash/prices?limit=1'
Responses
| Status | Meaning | Body type |
|---|---|---|
200 | Incrementally prepared local data | ModelPricePage |
400 | Invalid query | ErrorResponse |
404 | Not present in the prepared catalog | ErrorResponse |
503 | Catalog preparation has not completed yet | ErrorResponse |
Response example · 200
Illustrative values; schema-valid and not a live catalog measurement.
{
"data": [
{
"source_id": "openRouter",
"platform": "OpenRouter",
"source_model_id": "google/gemini-2.5-flash",
"unit": "input_tokens_per_million",
"amount_usd": 0.3,
"currency": "USD"
}
],
"platform_profiles": [
{
"source_model_id": "google/gemini-2.5-flash",
"publisher": "example",
"architecture": {
"input_modalities": [
"example"
],
"output_modalities": [
"example"
]
},
"hosting": "example",
"source_id": "openRouter",
"platforms": [
"example"
],
"name": "Google: Gemini 2.5 Flash",
"description": "Illustrative model observation; consult live evidence for current specifications.",
"context_length": 1048576
}
],
"total_count": 1,
"limit": 1,
"offset": 0
}
| Field | Type | Required | Meaning / constraints |
|---|---|---|---|
data | array of PricePointDto | Yes | Max. items: 100 |
platform_profiles | array of PlatformProfile | Yes | |
total_count | integer | Yes | Minimum: 0 |
limit | integer | Yes | Minimum: 1; Maximum: 100 |
offset | integer | Yes | Minimum: 0 |