Sections12
Reading time14 min
EvidenceEngineering reference

Bring your own key to any of these providers

Google Gemini logoGoogle Gemini
Mistral logoMistral
Anthropic logoAnthropic
OpenAI logoOpenAI
Groq logoGroq
Cloudflare logoCloudflare
NVIDIA logoNVIDIA
Cerebras logoCerebras
OpenRouter logoOpenRouter
OVHcloud logoOVHcloud
GitHub Models logoGitHub Models
Z.AI (Zhipu) logoZ.AI (Zhipu)
Hugging Face logoHugging Face
SambaNova logoSambaNova
Together AI logoTogether AI
Moonshot (Kimi) logoMoonshot (Kimi)
Fireworks AI logoFireworks AI
Cohere logoCohere
Azure OpenAI logoAzure OpenAI
Amazon Bedrock logoAmazon Bedrock
xAI (Grok) logoxAI (Grok)
Google Vertex AI logoGoogle Vertex AI
Custom endpoint logoCustom endpoint
Ollama logoOllama
Google Gemini logoGoogle Gemini
Mistral logoMistral
Anthropic logoAnthropic
OpenAI logoOpenAI
Groq logoGroq
Cloudflare logoCloudflare
NVIDIA logoNVIDIA
Cerebras logoCerebras
OpenRouter logoOpenRouter
OVHcloud logoOVHcloud
GitHub Models logoGitHub Models
Z.AI (Zhipu) logoZ.AI (Zhipu)
Hugging Face logoHugging Face
SambaNova logoSambaNova
Together AI logoTogether AI
Moonshot (Kimi) logoMoonshot (Kimi)
Fireworks AI logoFireworks AI
Cohere logoCohere
Azure OpenAI logoAzure OpenAI
Amazon Bedrock logoAmazon Bedrock
xAI (Grok) logoxAI (Grok)
Google Vertex AI logoGoogle Vertex AI
Custom endpoint logoCustom endpoint
Ollama logoOllama

Compare hosted, own-key, and local processing

The review workspace stays consistent, but account ownership, credentials, provider terms, model availability, and PE treatment differ by route. Check those differences before processing a sensitive document or scheduling a batch.

Zero Setup

Hosted processing

Dynamite Docs manages model discovery and routing. Extractions draw directly from your plan allowance.

Direct Wholesale

Own provider key

Bring verified keys from Google, Groq, Mistral, OpenAI, Anthropic, or custom gateways. AES-256-GCM encrypted.

100% On-Device

Local companion

Inference stays on your physical machine via authenticated loopback on port 8756. Zero external data transmission.

DimensionHosted processingOwn provider keyLocal companion
Model accountManaged by Dynamite DocsYour verified provider accountYour local Ollama installation
Request routeEligible hosted providerProvider endpoint for the selected keySigned loopback connection on port 8756
CredentialsNo provider key to configureKey encrypted with AES-256-GCMSigned local token required for every route
Model availabilityDiscovered by the hosted routerDiscovered from the connected accountModels installed in local Ollama
Plan accountingSuccessful pages use hosted PEFree and Starter use PE; Pro and Scale cost 0 PEEligibility and usage appear in the workspace
Review and exportSource beside proposed fields and rowsSame source-first review workspaceSame source-first review workspace
AI document extraction provider dashboard
Compare hosted document AI, connected provider keys, local processing, and the remaining page allowance before extraction.

Provider billing, limits, and data handling

The provider account, model availability, API limits, and provider charges remain under your control.

Scans and photos need a vision-capable model. A text-only model cannot inspect the pixels on an image-only page.

Provider retention, training, and regional terms differ. Check the selected route before sending a sensitive document.

A model can return an incomplete or incorrect value. The source-first review and export steps stay the same whichever route you select.

How BYOK document extraction works inside Dynamite Docs

Connect your API key in settings

Open Settings > AI & Processing in your workspace. Choose your provider from Google Gemini, Groq, Mistral, OpenAI, Anthropic, OpenRouter, or Custom Endpoint. Paste your secret key and click Verify & Save Key. The server runs an instant health check to discover available models and encrypts the key at rest using AES-256-GCM.

Configure active models and fallback routing

Choose from the models discovered for the verified key. If you configure a fallback, make sure it also supports the document type and satisfies the same privacy policy. A fallback reduces dependency on one route but does not guarantee availability.

Set data privacy and residency policies

Use the policy controls to exclude routes that do not meet the active training or residency rule. When cloud transmission is not acceptable, use the signed local Ollama companion on an eligible plan.

Extract, review, and export

Upload a supported document and run it with the selected route. Compare proposed fields and rows with the source before export. Free and Starter own-key pages use the monthly PE allowance; Pro and Scale own-key runs cost 0 PE. Provider charges and limits still apply.

How to connect your AI provider inside Dynamite Docs

Follow these step-by-step instructions to generate your API keys, connect credentials in Settings > AI & Processing, and configure active vision models for automated extraction.

Google AI Studio (Gemini) logo

Google AI Studio (Gemini)

Wholesale: ~$0.0002 / page

Google AI Studio gives developers direct access to Gemini 2.0 Flash and 1.5 Pro. It delivers sub-second extraction on invoices and receipts, plus a 2,000,000 token context window that parses 150-page statement batches in a single prompt without row loss across page boundaries.

Connection instructions inside Dynamite Docs

STEP 01
Generate key in Google AI Studio

Sign in at aistudio.google.com/app/apikey with your Google profile, click "Create API key", and link to an existing Google Cloud project with billing enabled or initialize a new project.

STEP 02
Navigate to AI Settings in Dynamite Docs

In your Dynamite Docs workspace, open Settings > AI & Processing (/app). Choose "Google Gemini" from the provider catalog dropdown.

STEP 03
Paste key and run verification

Paste your AIzaSy... key and click "Verify & Save Key". The server performs an encrypted validation ping to ensure your token is active and discovers available models.

STEP 04
Set active and fallback models

Select gemini-2.0-flash as your primary model for high-velocity intake. Configure gemini-1.5-pro for complex multi-page financial statements.

Technical specifications & pricing parameters
Recommended Modelgemini-2.0-flash
Token Pricing$0.10 / 1M input tokens (~$0.0002 per invoice page)
Extraction Latency1.2s to 1.8s per document
Context Window2,000,000 tokens (up to 150 pages)
Zero-Training TermsCommercial developer inputs not used to train models
Google Gemini API key setup in Dynamite Docs
Paste your Google AI Studio key, verify connectivity, and inspect discovered vision models.
Dynamite Docs logoDynamite Docs
The handoff
Processing settings: Your model, your accountConnect your API key in settings. Configure active models and fallback routing. Set data privacy and residency policies. Extract, review, and export. Illustrative example, not extracted customer data.CHOOSE HOW TO PROCESS THE FILEHosted AIUse the hosted allowanceYour provider keyChoose a provider and modelLocal OllamaOn an eligible planSamesource-linked…Allowances and availability depend on your plan and provider.
  1. 01Connect your API key in settings
  2. 02Configure active models and fallback routing
  3. 03Set data privacy and residency policies
  4. 04Extract, review, and export
Choose the provider for each run. Illustrative example.

Supported providers, endpoints, and architectural controls

The available model list is discovered from each verified account. Provider keys are encrypted with AES-256-GCM before storage:

Google Gemini logo
Google GeminiUse models returned by the connected Google account and check vision support.
Groq Cloud logo
Groq CloudUse models returned by the connected Groq account and its current limits.
Mistral AI logo
Mistral AIUse models returned by the connected Mistral account and its regional terms.
OpenAI & Anthropic logoOpenAI & Anthropic logo
OpenAI & AnthropicChoose a vision-capable model for scans and images; check current API terms.
Cloud Platforms logo
Cloud PlatformsAzure OpenAI, Amazon Bedrock, Google Vertex AI, and OpenRouter gateways
OpenAI-Compatible Endpoints logo
OpenAI-Compatible EndpointsLiteLLM, vLLM, Ollama server, or internal private VPC endpoints
Local Ollama Companion logo
Local Ollama CompanionSigned loopback connection on port 8756 for model inference on your computer.
PE treatmentFree and Starter own-key pages use PE; Pro and Scale own-key runs cost 0 PE.

24 supported providers

Router

1

Local

1

Self-host

1

Plus any OpenAI-compatible endpoint: a custom base URL and model id route straight to your own gateway. Keys are encrypted with AES-256-GCM and used only to route your own requests.

Compare AI OCR and document extraction models
Filter models by provider and capability, then choose a vision or text model that matches the source document.

A realistic example

Dynamite Docs logoDynamite Docs
Inside the output
Processing settings: Your model, your accountGoogle Gemini: Use models returned by the connected Google account and check vision support.. Groq Cloud: Use models returned by the connected Groq account and its current limits.. Mistral AI: Use models returned by the connected Mistral account and its regional terms.. Illustrative example, not extracted customer data.Source-linked field inspectorProcessing settingsUse models returned by the…Use models returned by the…Use models returned by the…Google GeminiUse models returned by theconnected Google account…Groq CloudUse models returned by theconnected Groq account and…Mistral AIUse models returned by theconnected Mistral account…
Google Gemini
Use models returned by the connected Google account and check vision support.
Groq Cloud
Use models returned by the connected Groq account and its current limits.
Mistral AI
Use models returned by the connected Mistral account and its regional terms.
Choose the provider for each run. Illustrative example.

Possible routes for a mixed batch of accounting documents:

Document
Clean digital supplier invoice (PDF)Hosted or own-key modelThe file has selectable text, but fields and line-item structure still need review.
Crumpled store receipt photo (JPG)Vision-capable hosted or own-key modelThe model must inspect pixels, layout, and small receipt text.
60-page historical bank statement (Scan)Vision-capable provider with suitable limitsCheck the provider context and request limits, then review page joins and running balances.
Confidential executive agreement (PDF)Local Ollama companionModel inference runs through the signed loopback connection instead of an external model API.

Developer API and cURL integration

Connect provider keys and execute extractions programmatically. Automated endpoints authenticate with your scoped API token and route requests through your connected models.

Verify and store an encrypted provider key

Connect an external API key. The server validates credentials against the provider API, discovers available models, and saves the secret encrypted with AES-256-GCM.

cURL RequestPOST /api/ai/keys
curl -X POST \
  -H "Authorization: Bearer <clerk_token>" \
  -H "Content-Type: application/json" \
  -d '{
    "provider": "groq",
    "apiKey": "gsk_a8b9c0d1e2f3...",
    "preferredModel": "<discovered_model_id>"
  }' \
  "https://app.dynamitedocs.com/api/ai/keys"
JSON Response (Status: 200 OK)
{
  "success": true,
  "provider": "groq",
  "status": "connected",
  "discoveredModels": [
    "<discovered_model_id>",
    "<another_available_model_id>"
  ],
  "activeModel": "<discovered_model_id>",
  "verifiedAt": "2026-09-15T15:00:00.000Z"
}

Model-agnostic human review and reconciliation

Source-first review

Test the output again when you change models

A different model can return different fields, column names, or values. Dynamite Docs keeps the source, editable result, confidence cues, and export controls in the same workspace, so the reviewer can compare the first result from a new route before it reaches a spreadsheet or downstream system.

Inline Document Correction

Inspect bounding boxes and override values directly beside the original source before export.

Pattern Memory & Field Learning

User corrections are remembered for the document pattern so subsequent runs inherit the corrected schema.

Side-by-side document review with confidence scores in Dynamite Docs
Inspect extracted line items alongside the original document page, review confidence highlights, and correct values before export.

$0

markup on inference

With your own key you pay your provider's published rate. We never add a reseller margin to a BYOK request.

Unlimited

BYOK on Pro and Scale

Own-key processing never counts against the hosted monthly allowance. You pay the provider directly.

24

auto-registered providers

Discovered live at startup, not hardcoded. Plus any OpenAI-compatible endpoint and a local Ollama companion.

A run counts as BYOK only when it uses your provider API key. Runs that use Dynamite Docs model keys count as hosted processing. BYOK on Free and Starter uses the monthly PE allowance; BYOK on Pro and Scale is unlimited.

What provider routing does and does not do

Private document AI processing controls
Restrict sensitive documents to approved providers, no-training routes, or local processing before extraction starts.

Zero-Training Guarantee

Sensitive mode blocks non-compliant routes. Inputs are never retained for model weights.

AES-256-GCM Encryption

Credentials are encrypted before disk write and never exposed or leaked in logs.

Data Residency & Local

Enforce EU data sovereignty or switch to local Ollama on loopback port 8756.

Does

  • Sends an authenticated extraction request to the hosted, own-key, custom, or local route you select.
  • Encrypts stored provider keys with AES-256-GCM.
  • Discovers the models currently available to a verified provider account.
  • Applies configured provider-policy and fallback rules before an eligible request is routed.
  • Supports model inference through the signed local Ollama companion on an eligible plan.

Does not

  • Does not control a provider's prices, model availability, rate limits, or account terms.
  • Does not make a text-only model capable of reading an image-only document.
  • Does not guarantee that a fallback route is available or policy-compatible.
  • Does not make a model response correct without source review.
  • Does not bypass Free or Starter PE limits when you use your own key.

Keep the review contract stable when the model changes

Provider models and account limits change. The useful constant is the workflow around them: a supported source file, proposed fields and rows, source review, corrections, and an explicit export.

Changing the model does not prove that its output matches the last model. Keep required columns and validation checks in the workspace, then compare high-risk values with the source before delivery.

Choose a hosted, own-key, custom, or local route according to the document type, provider terms, and the controls your organization requires.

Provider setup guide11 min read · Updated October 2026

Use Your Own API Key for AI Document Extraction

Follow the connection steps for supported provider accounts and local Ollama. The guide also covers model discovery, policy checks, rate limits, source review, and a representative test before a batch run.

Questions, answered

Related integrations

Document workflows it feeds

From the blog

Your key, your model, your rate

Connect a provider and test one representative document.

Open Settings > AI & Processing, verify the key, choose a discovered model, and compare the first extraction with the source before using that route for a batch.

Can we save optional cookies?

Essential cookies keep you signed in. With your permission, optional analytics cookies and masked session replays help us understand usage and diagnose errors. Privacy policy. Change your choice anytime in Cookie settings.