What is the AI API Pricing Dataset?

It is a machine-readable dataset of public AI API model pricing records normalized for model comparison, API cost estimation, and budget planning.

What does a null price mean?

Null means the value is not listed, not applicable, not published, or not currently available in the dataset. It is not a zero-dollar price.

Are these official billing numbers?

No. The dataset links to official provider pricing sources, but contracts, invoices, and provider billing dashboards remain the final source for billing decisions.

Dataset

Verified AI API & LLM Pricing Dataset

Name: AI API Pricing Dataset
Creator: AICostBudget
License: https://creativecommons.org/licenses/by/4.0/

A source-linked AI API pricing dataset covering token, cache, batch, tiered, multimodal, speech, and non-token pricing across multiple providers. Live JSON and CSV include official source links and verification metadata; the fixed v1.1.0 release is available for reproducible research. Provider contracts and invoices remain the final billing authority.

Use JSON Download CSV View schema

CC BY 4.0Last verified Oct 15, 2026Cite this dataset

Last verifiedOct 15, 2026

Providers covered11

Pricing records89

Schema1.10.0

Live data version1.0.0

FormatsJSON + CSV

Statistics count public pricing records returned by the public pricing adapter with Pricing V2 enabled. The verification date comes from synced pricing metadata, not page build time.

Dataset resources

Live data and reproducible references.

AICostBudget maintains source-linked AI API pricing records. The live export can change; the archived release and DOI identify a fixed snapshot.

Canonical live datasetCurrent documentation and JSON/CSV access GitHub sourceRepository and validation history Hugging FacePricing V2 dataset viewer Fixed v1.1.0 releaseSnapshot 2026-09-30 Zenodo version DOI10.5281/zenodo.23087250 MethodologyNormalization, sources, and limitations

Live dataset

Use the current Website JSON/CSV for operational comparisons and cite this page with your access date. Records carry official provider pricing or documentation URLs and verification metadata where available; last_verified_at records verification, while checked_at records a source check.

Frozen release

Use v1.1.0 for reproducible research or a fixed historical reference. It captures the 2026-09-30 snapshot at dataset commit 6882e95a8cdf5fb39b75b47bcd9ba0bff63b13d8, archived under DOI 10.5281/zenodo.23087250. It is not a promise that today's live prices match that snapshot.

Developer quickstart

Use the dataset in 60 seconds.

Fetch the public JSON dataset without an SDK, or download CSV for spreadsheets and data imports. No managed API service is implied.

Public JSON dataset endpointhttps://aicostbudget.com/api/datasets/ai-api-pricing.json

JavaScript

const data = await fetch("https://aicostbudget.com/api/datasets/ai-api-pricing.json").then((res) => res.json());
const model = data.records.find((row) =>
  row.provider_id === "openai" && row.model_id === "gpt-6-astra"
);
console.log(model.model, model.pricing_components);

Python standard library

import json
from urllib.request import urlopen

with urlopen("https://aicostbudget.com/api/datasets/ai-api-pricing.json") as response:
    data = json.load(response)
model = next(row for row in data["records"] if
    row["provider_id"] == "openai" and row["model_id"] == "gpt-6-astra")
print(model["model"], model["pricing_components"])

Prefer CSV? Download one row per model. Structured arrays are preserved as JSON strings in fields such as pricing_tiers_json and pricing_components_json.

Why this dataset is different

Pricing contracts, not just headline token rates.

Each exported record keeps the evidence and billing structure needed to interpret the price safely.

Source-linked verification

Records retain official source URLs, verification status, checked dates, and last-verified dates.

Tier-aware pricing

Context thresholds and whole-request pricing rules remain structured instead of being collapsed into one rate.

Conditional components

Cache writes, processing modes, effective periods, regional rules, and allowances can remain distinct components.

Native non-token units

Document pages, storage, requests, and other supported dimensions are not mislabeled as token prices.

Source verification proof: 23 distinct official source URLs are represented in the current export. Last verified Oct 15, 2026; provider contracts and invoices remain the final billing authority. Review official sources.

Access and distribution

Choose the format and destination for your workflow.

Website JSON and CSV expose the public Pricing V2 projection used on this page and mirrored on Hugging Face. GitHub Pages /api/v1 remains a separate compatibility projection and may contain a different record count by design.

Use JSON Download CSV View schema

Dataset identity

Live data version 1.0.0; public export schema 1.10.0; USD; multiple billing units. This live version is separate from the frozen v1.1.0 release. Null values mean not listed, not published, not applicable, or not currently available in the dataset.

Schema compatibility

Schema 1.3.0 adds pricing components without removing or renaming existing fields. Schema 1.4.0 adds structured request-time pricing for models with time-dependent rates. Schema 1.5.0 adds explicit non-token billing fields. Schema 1.6.0 adds generic usage-tier conditions. Schema 1.7.0 adds eligibility-scoped conditional usage allowances without mixing them into standard paid tiers. Schema 1.8.0 adds transport, modality-aware alternatives, optional feature charges, and model-selection metadata.

Schema 1.10.0 adds optional structured promotion and list-price conditions. JSON and CSV retain current promotional charges, list charges, provider-stated duration, and unknown end dates without changing existing CSV columns.

Schema 1.9.0 adds authoritative model access and identity-binding metadata to public JSON and appends the corresponding CSV columns. This additive extension preserves existing field names, order, and semantics. JSON consumers that ignore unknown fields remain backward compatible; CSV consumers relying on fixed column counts must accept appended columns.

Access metadata is separate from pricing verification. Approved binding means the identity-binding review passed; it does not guarantee API access for a user account. Exporting this metadata does not enable access enforcement.

License: Creative Commons Attribution 4.0

Website

Canonical documentation, schema, methodology, citation, and developer quickstart.

JSON / CSV

Read-only Website endpoints for the full public Pricing V2 projection used by this page.

JSON CSV

Price change feed

Atom entries for verified price-relevant model changes, with stable anchors and official-source links.

Open Atom feed

GitHub

Source, versioning, raw artifacts, methodology, validation workflows, and change history.

Open repository

Hugging Face

Dataset discovery, files, and the tabular viewer for the Website projection.

Open dataset

GitHub Pages

A read-only static V1 dataset artifact from the repository. Its intentionally narrower schema is not the Website Pricing V2 export.

Open static JSON

Want to compare model prices visually instead? Use the Model Pricing Comparison. This Dataset page is for machine-readable pricing used in code, spreadsheets, calculators, routing analysis, budget systems, and internal cost tooling.

Coverage

Provider coverage and supported pricing fields.

Counts are derived from public pricing rows. Provider cards link to the matching API pricing pages instead of repeating every model row here.

OpenAI

OpenAI API Pricing

Records: 19
Input rows: 18
Output rows: 18
Cached input: Available
Batch fields: Available

View provider pricingAnthropic

Anthropic API Pricing

Records: 14
Input rows: 14
Output rows: 14
Cached input: Available
Batch fields: Available

View provider pricingGoogle Gemini

Google Gemini API Pricing

Records: 18
Input rows: 16
Output rows: 16
Cached input: Available
Batch fields: Available

View provider pricingxAI

xAI API Pricing

Records: 13
Input rows: 9
Output rows: 9
Cached input: Available
Batch fields: Available

View provider pricingDeepSeek

DeepSeek API Pricing

Records: 4
Input rows: 3
Output rows: 3
Cached input: Available
Batch fields: Not currently available

View provider pricingMistral AI

Mistral AI API Pricing

Records: 10
Input rows: 8
Output rows: 8
Cached input: Available
Batch fields: Available

View provider pricingCohere

Cohere API Pricing

Records: 5
Input rows: 2
Output rows: 2
Cached input: Not currently available
Batch fields: Not currently available

View provider pricingMoonshot AI

Moonshot AI API Pricing

Records: 3
Input rows: 3
Output rows: 3
Cached input: Available
Batch fields: Available

View provider pricing

Google Cloud

Google Cloud Document AI pricing data

Records: 1
Input rows: 0
Output rows: 0
Cached input: Not currently available
Batch fields: Not currently available

Dataset and API coverage

Amazon Web Services

Amazon Textract pricing data

Records: 1
Input rows: 0
Output rows: 0
Cached input: Not currently available
Batch fields: Not currently available

Dataset and API coverage

Microsoft Azure

Azure Document Intelligence pricing data

Records: 1
Input rows: 0
Output rows: 0
Cached input: Not currently available
Batch fields: Not currently available

Dataset and API coverage

Non-token pricing

Flat and structured non-token pricing records.

Flat records show a native unit price. Tiered, regional, and conditional records preserve their component-level pricing rules without being reduced to one scalar.

Amazon Web Services

Amazon Textract DetectDocumentText

Structured pricing · Tiered · Regional · Conditional offer · Free allowance · 2 pricing components

Model ID: DetectDocumentText
Lifecycle: active
Modality: Not listed
Verification: verified · 2026-10-09

View structured pricing for Amazon Textract DetectDocumentText

Up to 1,000,000 pages

$1.50 / 1,000 pages

Component: document page
Modality: Document
Processing mode: Standard
Billing unit: per_1000_pages
Effective period: Effective date not published
Calculation default: No
Verification: Verified
Tier mode: Graduated
Region: Region-specific pricing; 16 endpoint geographies and 1 data residency
Regional adjustment: us-west-1: multiply by 1.4
Regional adjustment: us-gov-east-1, us-gov-west-1: multiply by 1.26

Above 1,000,000 pages

$0.60 / 1,000 pages

Component: document page
Modality: Document
Processing mode: Standard
Billing unit: per_1000_pages
Effective period: Effective date not published
Calculation default: No
Verification: Verified
Tier mode: Graduated
Region: Region-specific pricing; 16 endpoint geographies and 1 data residency
Regional adjustment: us-west-1: multiply by 1.4
Regional adjustment: us-gov-east-1, us-gov-west-1: multiply by 1.26

Conditional usage allowance

1,000 pages per month for 3 months.

Eligibility: New customers on paid accounts only.

Official pricing source

azure

Azure Document Intelligence Read

Structured pricing · Tiered · Regional · 2 pricing components

Model ID: prebuilt-read
Lifecycle: active
Modality: Not listed
Verification: verified · 2026-10-09

View structured pricing for Azure Document Intelligence Read

Up to 1,000,000 pages

$1.50 / 1,000 pages

Component: document page
Modality: Document
Processing mode: Standard
Billing unit: per_1000_pages
Effective period: Effective date not published
Calculation default: No
Verification: Verified
Tier mode: Graduated
Region: Region-specific pricing; 31 endpoint geographies and 1 data residency

Above 1,000,000 pages

$0.60 / 1,000 pages

Component: document page
Modality: Document
Processing mode: Standard
Billing unit: per_1000_pages
Effective period: Effective date not published
Calculation default: No
Verification: Verified
Tier mode: Graduated
Region: Region-specific pricing; 31 endpoint geographies and 1 data residency

Official pricing source

Cohere

Cohere Parse v5

$1.50 / 1,000 pages

Model ID: parse-v5.0
Billing unit: per_1000_pages
Pricing dimension: document_page
Token fields: null (not applicable)

Official pricing source

Google Cloud

Enterprise Document OCR

Structured pricing · Tiered · Regional · Free allowance · 3 pricing components

Model ID: OCR_PROCESSOR
Lifecycle: active
Modality: Not listed
Verification: verified · 2026-10-09

View structured pricing for Enterprise Document OCR

First 1,000 pages

$0.00 / 1,000 pages

Component: document page
Modality: Document
Processing mode: Standard
Billing unit: per_1000_pages
Effective period: Effective date not published
Calculation default: No
Verification: Verified
Tier mode: Graduated
Region: Global base pricing; 8 endpoint geographies and 1 data residency

1,000-5,000,000 pages

$1.50 / 1,000 pages

Component: document page
Modality: Document
Processing mode: Standard
Billing unit: per_1000_pages
Effective period: Effective date not published
Calculation default: No
Verification: Verified
Tier mode: Graduated
Region: Global base pricing; 8 endpoint geographies and 1 data residency

Above 5,000,000 pages

$0.60 / 1,000 pages

Component: document page
Modality: Document
Processing mode: Standard
Billing unit: per_1000_pages
Effective period: Effective date not published
Calculation default: No
Verification: Verified
Tier mode: Graduated
Region: Global base pricing; 8 endpoint geographies and 1 data residency

Official pricing source

Google

Gemini 3.8 Flash

Structured pricing · Regional · 8 pricing components

Model ID: gemini-3.8-flash
Lifecycle: Standard
Modality: Text → Text
Verification: verified · 2026-10-09

View structured pricing for Gemini 3.8 Flash

Batch · Storage · 2026-09-02 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Batch
Billing unit: per_1m_tokens_per_hour
Effective period: 2026-09-02 through 2026-12-31
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Batch · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Batch
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Flex · Storage · Effective date not published

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Flex
Billing unit: per_1m_tokens_per_hour
Effective period: Effective date not published
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Flex · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Flex
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Priority · Storage · Effective date not published

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Priority
Billing unit: per_1m_tokens_per_hour
Effective period: Effective date not published
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Priority · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Priority
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Storage · Current default · 2026-09-02 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Standard
Billing unit: per_1m_tokens_per_hour
Effective period: Current default · 2026-09-02 through 2026-12-31
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Standard
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

Google

Gemini 3.8 Flash-Lite TTS

Structured pricing · Regional · 8 pricing components

Model ID: gemini-3.8-flash-lite-tts
Lifecycle: active
Modality: Text → Audio
Verification: verified · 2026-09-29

View structured pricing for Gemini 3.8 Flash-Lite TTS

Batch · Storage · 2026-09-22 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Batch
Billing unit: per_1m_tokens_per_hour
Effective period: 2026-09-22 through 2026-12-31
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Batch · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Batch
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Flex · Storage · 2026-09-22 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Flex
Billing unit: per_1m_tokens_per_hour
Effective period: 2026-09-22 through 2026-12-31
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Flex · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Flex
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Priority · Storage · 2026-09-22 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Priority
Billing unit: per_1m_tokens_per_hour
Effective period: 2026-09-22 through 2026-12-31
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Priority · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Priority
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Storage · Current default · 2026-09-22 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Standard
Billing unit: per_1m_tokens_per_hour
Effective period: Current default · 2026-09-22 through 2026-12-31
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Standard
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

Google

Gemini 3.8 Flash TTS

Structured pricing · Regional · 8 pricing components

Model ID: gemini-3.8-flash-tts
Lifecycle: active
Modality: Text → Audio
Verification: verified · 2026-09-29

View structured pricing for Gemini 3.8 Flash TTS

Batch · Storage · 2026-09-22 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Batch
Billing unit: per_1m_tokens_per_hour
Effective period: 2026-09-22 through 2026-12-31
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Batch · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Batch
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Flex · Storage · 2026-09-22 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Flex
Billing unit: per_1m_tokens_per_hour
Effective period: 2026-09-22 through 2026-12-31
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Flex · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Flex
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Priority · Storage · 2026-09-22 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Priority
Billing unit: per_1m_tokens_per_hour
Effective period: 2026-09-22 through 2026-12-31
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Priority · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Priority
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Storage · Current default · 2026-09-22 through 2026-12-31

$0.50 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Standard
Billing unit: per_1m_tokens_per_hour
Effective period: Current default · 2026-09-22 through 2026-12-31
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Storage · Effective from 2027-01-01

$1.00 / 1m tokens per hour

Component: storage
Modality: Text
Processing mode: Standard
Billing unit: per_1m_tokens_per_hour
Effective period: Effective from 2027-01-01
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

Google

Gemini 3.8 Live

Structured pricing · Regional · 4 pricing components

Model ID: gemini-3.8-live
Lifecycle: active
Modality: Audio / Image / Text / Video → Text / Audio
Verification: verified · 2026-09-21

View structured pricing for Gemini 3.8 Live

Standard · Input · Current default · effective from 2026-09-15

$0.005 / minute

Component: input
Modality: Audio
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-09-15
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Input · Current default · effective from 2026-09-15

$0.002 / minute

Component: input
Modality: Image
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-09-15
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Input · Current default · effective from 2026-09-15

$0.002 / minute

Component: input
Modality: Video
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-09-15
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Output · Current default · effective from 2026-09-15

$0.018 / minute

Component: output
Modality: Audio
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-09-15
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

Google

Gemini 3.8 Live Extended Thinking

Structured pricing · Regional · 4 pricing components

Model ID: gemini-3.8-live-extended-thinking
Lifecycle: active
Modality: Audio / Image / Text / Video → Text / Audio
Verification: verified · 2026-09-21

View structured pricing for Gemini 3.8 Live Extended Thinking

Standard · Input · Current default · effective from 2026-09-15

$0.005 / minute

Component: input
Modality: Audio
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-09-15
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Input · Current default · effective from 2026-09-15

$0.002 / minute

Component: input
Modality: Image
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-09-15
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Input · Current default · effective from 2026-09-15

$0.002 / minute

Component: input
Modality: Video
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-09-15
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Output · Current default · effective from 2026-09-15

$0.018 / minute

Component: output
Modality: Audio
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-09-15
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

Mistral AI

Mistral OCR 4.0

$4.00 / 1,000 pages

Model ID: mistral-ocr-4-0
Billing unit: per_1000_pages
Pricing dimension: document_page
Token fields: null (not applicable)

Official pricing source

Mistral AI

Mistral OCR 4.1

$4.00 / 1,000 pages

Model ID: mistral-ocr-4-1
Billing unit: per_1000_pages
Pricing dimension: document_page
Token fields: null (not applicable)

Official pricing source

OpenAI

GPT-Live 1

Structured pricing · Regional · 1 pricing component

Model ID: gpt-live-1
Lifecycle: active
Modality: Not listed
Verification: verified · 2026-10-09

View structured pricing for GPT-Live 1

Standard · Session duration · Current default · effective from 2026-09-10

$0.05 / minute

Component: session duration
Modality: Audio
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-09-10
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

xAI

Grok Voice 2.0

Structured pricing · Regional · 2 pricing components

Model ID: grok-voice-2.0
Lifecycle: deprecated
Modality: Text → Not listed
Verification: verified · 2026-10-15

View structured pricing for Grok Voice 2.0

Standard · Input · Current default · effective from 2026-07-29

$0.004 / text input

Component: input
Modality: Text
Processing mode: Standard
Billing unit: per_text_input
Effective period: Current default · effective from 2026-07-29
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Session duration · Current default · effective from 2026-07-29

$0.08 / minute

Component: session duration
Modality: Audio
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-07-29
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

xAI

Grok Voice 3.0

Structured pricing · Regional · 2 pricing components

Model ID: grok-voice-3.0
Lifecycle: active
Modality: Text → Not listed
Verification: verified · 2026-10-15

View structured pricing for Grok Voice 3.0

Standard · Input · Current default · effective from 2026-10-14

$0.004 / text input

Component: input
Modality: Text
Processing mode: Standard
Billing unit: per_text_input
Effective period: Current default · effective from 2026-10-14
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Session duration · Current default · effective from 2026-10-14

$0.08 / minute

Component: session duration
Modality: Audio
Processing mode: Standard
Billing unit: per_minute
Effective period: Current default · effective from 2026-10-14
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

xAI

Grok Voice Transcribe 1.0

Structured pricing · Regional · 2 pricing components

Model ID: grok-voice-transcribe-1.0
Lifecycle: retired
Modality: Not listed
Verification: verified · 2026-09-29

View structured pricing for Grok Voice Transcribe 1.0

Standard · Session duration · Current default · effective from 2026-09-17

$0.10 / hour

Component: session duration
Modality: Audio
Processing mode: Standard
Billing unit: per_hour
Effective period: Current default · effective from 2026-09-17
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Session duration · Effective from 2026-09-17

$0.20 / hour

Component: session duration
Modality: Audio
Processing mode: Standard
Billing unit: per_hour
Effective period: Effective from 2026-09-17
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

xAI

Grok Voice Transcribe 2.0

Structured pricing · Regional · 2 pricing components

Model ID: grok-voice-transcribe-2.0
Lifecycle: active
Modality: Not listed
Verification: verified · 2026-09-29

View structured pricing for Grok Voice Transcribe 2.0

Standard · Session duration · Current default · effective from 2026-09-17

$0.10 / hour

Component: session duration
Modality: Audio
Processing mode: Standard
Billing unit: per_hour
Effective period: Current default · effective from 2026-09-17
Calculation default: Yes
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Standard · Session duration · Effective from 2026-09-17

$0.20 / hour

Component: session duration
Modality: Audio
Processing mode: Standard
Billing unit: per_hour
Effective period: Effective from 2026-09-17
Calculation default: No
Verification: Verified
Region: Global base pricing; 1 endpoint geography and 1 data residency

Official pricing source

Data Dictionary

Public fields in the JSON and CSV exports.

Existing scalar and tier fields remain aligned. JSON exposes full conditional pricing components; CSV stores the same array as valid JSON in its final column so each model remains one row.

Field	Format	Type	Meaning
`provider_id`	JSON and CSV	string	Stable provider identifier used by AICostBudget pricing pages.
`model_id`	JSON and CSV	string	Stable model identifier from the public pricing adapter.
`input_price_per_1m_tokens`	JSON and CSV	number or null	Published input token price in USD per 1M tokens. Null is not zero.
`cached_input_price_per_1m_tokens`	JSON and CSV	number or null	Compatibility/default cached-input or read-like scalar when available; provider cache semantics can differ.
`output_price_per_1m_tokens`	JSON and CSV	number or null	Published output token price in USD per 1M tokens. Null is not zero.
`unit_price`	JSON and CSV	number or null	Primary non-token unit price in USD; null for token-priced rows.
`billing_unit`	JSON and CSV	string or null	Machine-readable structured billing unit such as per_1000_pages or per_1m_tokens_per_hour.
`pricing_dimension`	JSON and CSV	string or null	Billable dimension such as document_page.
`pricing_tiers`	JSON	array	Structured canonical tier records with threshold, billing semantics, prices, source references, official URL, and verification time. Models without tiers use [].
`pricing_components`	JSON	array	Full conditional pricing structure: component, decimal-string amount, conditions (including generic usage_tier boundaries and free allowance), effective range, provenance, and verification status. Models without components use [].
`official_source_url`	JSON and CSV	string	Official provider pricing source URL saved with the pricing record.
`last_verified_at`	JSON and CSV	date string or null	Canonical verification time from the pricing projection. It is not page build time.

View all schema fields

Field	Format	Type	Meaning
`provider`	JSON and CSV	string	Provider display name.
`model`	JSON and CSV	string	Human-readable model name.
`currency`	JSON and CSV	string	Currency used by normalized pricing values. Current records use USD.
`pricing_unit`	JSON and CSV	string or null	Legacy text-token scalar unit. Structured multimodal and non-token records use null instead of mislabeling another modality or billing dimension as text-token pricing.
`billing_quantity`	JSON and CSV	number or null	Quantity represented by unit_price, such as 1000 pages.
`status`	JSON and CSV	string	Public model status from the pricing adapter, such as latest, legacy, deprecated, or specialized.
`availability`	JSON and CSV	string	Availability label from the pricing adapter.
`effective_from`	JSON and CSV	date string or null	Effective date when a selected price record exposes one.
`effective_until`	JSON and CSV	date string or null	End date when the selected price record exposes one.
`notes`	JSON and CSV	string or null	Adapter notes, warnings, or source context that should not be interpreted as another pricing field.
`pricing_tier_count`	JSON and CSV	integer	Number of canonical standard-pricing tiers for the model. Models without tiers use 0.
`pricing_tiers_json`	CSV	JSON string	The same pricing_tiers array serialized as valid JSON in one CSV cell, preserving one row per model.
`time_pricing`	JSON	object or null	Structured request-time pricing schedule for models whose rates depend on request time; null when no temporal pricing applies.
`time_pricing_json`	CSV	JSON string	Compact JSON serialization of time_pricing for CSV while preserving one row per model.
`pricing_components_json`	CSV	JSON string	Compact deterministic JSON serialization of pricing_components; [] when no components apply.
`model_selection`	JSON	object or null	Verified model-selection metadata, including default behavior and explicit model-ID support; null when not published.
`model_selection_json`	CSV	JSON string	Compact JSON serialization of model_selection; null when model-selection metadata is not published.

Methodology

How prices are normalized and verified.

Records are generated from the existing Pricing Table adapter with Pricing V2 enabled, so the page and downloads share the same public selector logic.

Token prices are normalized to USD per 1M tokens when the public provider source exposes compatible token pricing.

Non-token prices retain their native billable dimension and unit. They are published in structured fields and are not inserted into token-price scalars or calculators.

Input, cached input, and output prices remain separate fields. Missing public values are exported as null and shown as not currently available, not as 0.

Scalar prices remain the compatibility/default view. pricing_components is the detailed conditional view and can represent multiple cache-write durations, processing modes, context classes, or graduated page-usage tiers.

Batch pricing columns are not exported unless the current adapter exposes supported batch fields for public records.

Status, availability, source URLs, verification dates, effective dates, and notes are preserved from the adapter output where available.

Tier selection uses total prompt tokens, including both regular and cached input. A matching tier prices the whole request; the threshold is not progressive.

When providers change prices, the synced pricing projection and metadata must be updated before those changes appear on this page or in the downloads.

Official Sources

Provider pricing sources used by the dataset.

Links are grouped and deduplicated by provider. They are source references, not endorsements or sponsorship claims.

OpenAI

Anthropic

Google Gemini

xAI

DeepSeek

https://api-docs.deepseek.com/quick_start/pricing

Mistral AI

Cohere

Moonshot AI

Pricing Change Log

Current history artifact.

The history artifacts preserve canonical verification snapshots and structured change events when supported by source evidence. They should not be treated as a complete historical price ledger for every provider and model.

For a frozen research view, read the latest AI API Pricing Index (September 2026). It is frozen at the Sep 30, 2026 UTC cutoff, not the live pricing database; use this Dataset, provider pages, and Model Pricing Comparison for live pricing.

Browse recent verified price changes when you need a human-readable view of event dates, before-and-after rates, and official sources.

For an applied Anthropic workload decision, use verified cache pricing in the savings calculator.

For dated provider context behind the current snapshot, review the OpenAI API pricing change history. Official provider sources remain the final authority for billing decisions.

Found an outdated or incorrect price?

Report mismatches with the provider, model, official source URL, and the expected pricing field so the projection can be reviewed against public sources.

Contact AICostBudget

Tiered Pricing Example

Grok 4.20 (Non-Reasoning) uses threshold-based pricing from the official pricing URL. Pricing changes over time — data verified on 2026-10-09.

xAI · Grok 4.20 (Non-Reasoning) Verified

Last verified 2026-10-09

Short context

< 200K prompt tokens

Input (per 1M prompt tokens): $1.25
Cached input (per 1M prompt tokens): $0.20
Output (per 1M tokens): $2.50

Long context

≥ 200K prompt tokens

Input (per 1M prompt tokens): $2.50
Cached input (per 1M prompt tokens): $0.40
Output (per 1M tokens): $5.00

Pricing rule: Threshold-based (by prompt tokens)
Token threshold: 200,000 prompt tokens
Official source: https://docs.x.ai/developers/pricing

JSON preview (truncated)

{
  "model_id": "grok-4.20-0309-non-reasoning",
  "pricing_tier_count": 2,
  "pricing_tiers": [
    {
      "id": "price:xai/grok-4.20-0309-non-reasoning:standard:short:current",
      "label": "Short context",
      "context_class": "short",
      "prompt_token_threshold": 200000,
      "comparison": "less_than",
      "token_basis": "total_prompt_tokens",
      "currency": "USD",
      "pricing_unit": "per_1m_tokens",
      "input_price_per_1m_tokens": 1.25,
      "cached_input_price_per_1m_tokens": 0.2,
      "output_price_per_1m_tokens": 2.5,
      "verification_status": "verified"
    }
  ]
}

Showing the first pricing tier. Full canonical tier JSON is available through View JSON.

How to Cite

How to cite this dataset.

For reproducible work, cite the fixed v1.1.0 snapshot and its version DOI. For current operational use, cite the live dataset page with your access date. The frozen release is not the latest live export.

Plain text

AICostBudget. AICostBudget AI API Pricing Dataset, v1.1.0 (snapshot 2026-09-30). https://doi.org/10.5281/zenodo.23087250. GitHub Release: https://github.com/aicostbudget/ai-api-pricing-data/releases/tag/v1.1.0. Dataset commit: 6882e95a8cdf5fb39b75b47bcd9ba0bff63b13d8.

BibTeX

@misc{aicostbudget_2026_v110,
  author = {{AICostBudget}},
  title = {AICostBudget AI API Pricing Dataset},
  year = {2026},
  version = {1.1.0},
  doi = {10.5281/zenodo.23087250},
  url = {https://doi.org/10.5281/zenodo.23087250},
  note = {Snapshot 2026-09-30; Git commit 6882e95a8cdf5fb39b75b47bcd9ba0bff63b13d8}
}