Frequently Asked Questions

Common questions about Eden AI V3 API.

General

What is Eden AI V3?

Eden AI V3 is a unified AI API platform that provides:

Universal AI Endpoint - Single endpoint for all non-LLM features (text, OCR, image, translation)
OpenAI-Compatible LLM - Drop-in replacement for OpenAI’s chat completions API
Multi-Provider Support - Access 50+ AI providers through one interface
Persistent File Storage - Upload files once, use in multiple requests

Which endpoint should I use?

Choose based on your use case: Use /v3/llm/chat/completions for:

Conversational AI and chatbots
Text generation and completion
Vision/multimodal AI (analyzing images with LLMs)
Tool/function calling
Any use case requiring OpenAI-compatible format

Use /v3/universal-ai for:

Text analysis (sentiment, moderation, AI detection)
OCR and document parsing
Image generation and analysis
Text embeddings
Translation

See Getting Started for detailed endpoint comparison.

How does the model string format work?

V3 uses a model string to specify what feature and provider to use: Universal AI format:

feature/subfeature/provider[/model]

Examples:

text/moderation/openai
ocr/financial_parser/google
image/generation/openai/dall-e-3

LLM format (simplified):

provider/model

Examples:

openai/gpt-4
anthropic/claude-3-5-sonnet-20241022
google/gemini-pro

See Universal AI Getting Started and Chat Completions for details.

Authentication & Access

How do I get an API key?

Sign up at app.edenai.run
Navigate to your dashboard
Generate an API token under “API Keys”
Use it in the Authorization header: Bearer YOUR_API_KEY

See Authentication Guide for details.

Can I use my own provider API keys?

Yes! You can bypass Eden AI billing by providing your own provider API keys. This is useful for:

Using existing provider credits
Testing specific provider features
Cost optimization

Contact support or check your dashboard for instructions on adding custom provider keys.

What are the rate limits?

Rate limits vary by:

Your account tier
Provider being used
Specific feature

Default limits are displayed in your dashboard. Rate limit headers are included in API responses:

X-RateLimit-Limit - Total requests allowed
X-RateLimit-Remaining - Requests remaining
X-RateLimit-Reset - Time when limit resets

When rate limited, you’ll receive a 429 Too Many Requests response.

How much does it cost?

Pricing is pay-as-you-go based on:

Provider used
Feature/model called
Volume of data processed

Every API response includes a cost field showing the charge in USD for that request:

{
  "status": "success",
  "cost": 0.0015,
  "output": { ... }
}

See Monitor Usage and Costs for tracking and optimization.

File Handling

What file input methods are supported?

V3 supports three methods: 1. File Upload (recommended for reuse):

# Upload once
response = requests.post(
    "https://api.edenai.run/v3/upload",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    files={"file": open("document.pdf", "rb")}
)
file_id = response.json()["file_id"]

# Use multiple times
payload = {"model": "ocr/financial_parser/google", "input": {"file": file_id}}

2. File URL:

payload = {
    "model": "ocr/financial_parser/google",
    "input": {"file": "https://example.com/document.pdf"}
}

3. Base64 (inline):

payload = {
    "model": "text/moderation/openai",
    "input": {"text": "Sample text"}
}

See Upload Files for detailed guide.

How long are uploaded files stored?

Files uploaded to /v3/upload are stored for 7 days by default. After expiration:

Files are automatically deleted
File IDs become invalid
You’ll need to re-upload

The expires_at timestamp is included in upload responses:

{
  "file_id": "550e8400-e29b-41d4-a716-446655440000",
  "filename": "document.pdf",
  "bytes": 123456,
  "created_at": "2025-12-26T10:00:00Z",
  "expires_at": "2026-01-02T10:00:00Z"
}

What are the file size limits?

Limits vary by feature and provider:

Feature Type	Typical Limit	Notes
OCR	100 MB	Some providers support larger files
Image Analysis	20 MB	JPEG, PNG, WebP, GIF
LLM Vision	20 MB	Provider-dependent
Document Translation	50 MB	PDF, DOCX, TXT

Exceeding limits returns a 413 Payload Too Large error with specific limit information.

Which file formats are supported?

Images:

JPEG, PNG, WebP, GIF
TIFF (OCR only)

Documents:

PDF
DOCX, DOC
TXT, RTF

Audio:

MP3, WAV (provider-dependent)

Format support varies by provider. Check provider-specific documentation or use the API Discovery endpoint for details.

LLM Features

Why is streaming mandatory in V3?

All LLM responses in V3 use Server-Sent Events (SSE) streaming to:

Reduce perceived latency
Provide real-time token generation
Match OpenAI’s API behavior
Enable better UX for chat applications

Even when stream: false is sent, responses are streamed. See Streaming Responses for implementation guide.

How do I send images to LLMs?

V3 LLM supports multimodal inputs via message content arrays:

payload = {
    "model": "openai/gpt-4o",
    "messages": [
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "What's in this image?"},
                {
                    "type": "image_url",
                    "image_url": {
                        "url": "https://example.com/image.jpg"
                    }
                }
            ]
        }
    ]
}

Supports:

Image URLs
Base64 data URLs (data:image/jpeg;base64,...)
Uploaded file UUIDs

See Working with Media Files for complete guide.

Which providers support vision/multimodal?

Provider	Models	Image Support	File Support
OpenAI	gpt-4o, gpt-4-turbo	✓	✓
Anthropic	claude-3-opus, claude-3-5-sonnet	✓	✓
Google	gemini-1.5-pro, gemini-1.5-flash	✓	✓
Mistral	pixtral-12b	✓	-

See Vision Capabilities for provider comparison.

Does V3 support tool/function calling?

Yes! V3 supports OpenAI-compatible tool calling:

payload = {
    "model": "openai/gpt-4",
    "messages": [...],
    "tools": [
        {
            "type": "function",
            "function": {
                "name": "get_weather",
                "description": "Get current weather",
                "parameters": { ... }
            }
        }
    ]
}

Tool calling support varies by provider:

✓ OpenAI (all GPT-4+ models)
✓ Anthropic (Claude 3+)
✓ Google (Gemini 1.5+)
✓ Mistral (Large models)

Universal AI

How do I discover available features?

Use the built-in API discovery endpoints:

# List all features
response = requests.get(
    "https://api.edenai.run/v3/info",
    headers={"Authorization": "Bearer YOUR_API_KEY"}
)

# Get feature details
response = requests.get(
    "https://api.edenai.run/v3/info/text/moderation",
    headers={"Authorization": "Bearer YOUR_API_KEY"}
)
# Returns: providers, input schema, output schema

See Explore the API for details.

Can I use fallback providers?

Not directly in V3’s simplified model string format. However, you can implement fallback logic in your application:

def call_with_fallback(primary_model, fallback_model, input_data):
    try:
        return call_universal_ai(primary_model, input_data)
    except Exception as e:
        print(f"Primary failed: {e}, trying fallback...")
        return call_universal_ai(fallback_model, input_data)

result = call_with_fallback(
    "text/moderation/openai",
    "text/moderation/google",
    {"text": "Sample text"}
)

How do I optimize costs?

Best practices:

Choose cost-effective providers:
- Compare pricing in dashboard
- Check cost field in responses
Cache results when possible:
- Many features (embeddings, moderation) have deterministic outputs
- Store results for identical inputs
Use appropriate models:
- Don’t use premium models for simple tasks
- Match model capability to task complexity
Batch processing:
- Process multiple items in fewer API calls when supported
Monitor usage:
- Track costs via Monitor Usage
- Set up alerts in dashboard

I’ve been an Eden AI user for some time. Is the previous version going to disappear?

Not yet. We’ll continue supporting the previous version until the end of 2026. You can find everything here https://old-app.edenai.run Also, you can find the documentation here

Troubleshooting

401 Unauthorized

Cause: Invalid or missing API token Solutions:

Verify token is correct
Check Authorization header format: Bearer YOUR_API_KEY
Ensure token hasn’t been revoked
Generate new token in dashboard

# Correct format
headers = {
    "Authorization": "Bearer YOUR_API_KEY",
    "Content-Type": "application/json"
}

402 Payment Required - Insufficient Credits

Cause: Account has insufficient credits Solutions:

Add credits in dashboard
Check current balance
Review cost per request in responses

404 Not Found

Cause: Invalid endpoint or model string Solutions:

Verify endpoint URL is correct
Check model string format matches pattern
Use /v3/info to discover available features
Ensure provider supports requested feature

# Wrong
"model": "openai/text-moderation"  # ❌ Invalid format

# Correct
"model": "text/moderation/openai"  # ✓

422 Validation Error

Cause: Invalid request body or parameters Common issues:

Missing required fields
Invalid parameter types
File format not supported
Model string malformed

Solution: Check error response for specific field errors:

{
  "detail": [
    {
      "loc": ["body", "input", "text"],
      "msg": "field required",
      "type": "value_error.missing"
    }
  ]
}

429 Too Many Requests

Cause: Rate limit exceeded Solutions:

Implement exponential backoff
Check X-RateLimit-Reset header
Upgrade account tier for higher limits
Distribute requests over time

import time

def call_with_retry(url, payload, max_retries=3):
    for attempt in range(max_retries):
        response = requests.post(url, json=payload, headers=headers)

        if response.status_code == 429:
            retry_after = int(response.headers.get('Retry-After', 60))
            time.sleep(retry_after)
            continue

        return response.json()

    raise Exception("Max retries exceeded")

Invalid model string format

Cause: Model string doesn’t match expected pattern For Universal AI: Must be feature/subfeature/provider[/model]

# Wrong
"text-moderation-openai"  # ❌
"openai/text/moderation"  # ❌

# Correct
"text/moderation/openai"  # ✓
"text/moderation/openai/gpt-4"  # ✓

For LLM: Must be provider/model

# Wrong
"gpt-4"  # ❌
"openai"  # ❌

# Correct
"openai/gpt-4"  # ✓

Provider temporarily unavailable

Cause: Upstream provider experiencing issues Solutions:

Check Eden AI Status Page
Try alternative provider for same feature
Implement retry logic with exponential backoff
Use error response for specific provider error details

V3 Documentation

Faq

Frequently Asked Questions

General

What is Eden AI V3?

Which endpoint should I use?

How does the model string format work?

Authentication & Access

How do I get an API key?

Can I use my own provider API keys?

What are the rate limits?

How much does it cost?

File Handling

What file input methods are supported?

How long are uploaded files stored?

What are the file size limits?

Which file formats are supported?

LLM Features

Why is streaming mandatory in V3?

How do I send images to LLMs?

Which providers support vision/multimodal?

Does V3 support tool/function calling?

Universal AI

How do I discover available features?

Can I use fallback providers?

How do I optimize costs?

I’ve been an Eden AI user for some time. Is the previous version going to disappear?

Troubleshooting

401 Unauthorized

402 Payment Required - Insufficient Credits

404 Not Found

422 Validation Error

429 Too Many Requests

Invalid model string format

Provider temporarily unavailable

Next Steps

V3 Documentation

​Frequently Asked Questions

​General

​What is Eden AI V3?

​Which endpoint should I use?

​How does the model string format work?

​Authentication & Access

​How do I get an API key?

​Can I use my own provider API keys?

​What are the rate limits?

​How much does it cost?

​File Handling

​What file input methods are supported?

​How long are uploaded files stored?

​What are the file size limits?

​Which file formats are supported?

​LLM Features

​Why is streaming mandatory in V3?

​How do I send images to LLMs?

​Which providers support vision/multimodal?

​Does V3 support tool/function calling?

​Universal AI

​How do I discover available features?

​Can I use fallback providers?

​How do I optimize costs?

​I’ve been an Eden AI user for some time. Is the previous version going to disappear?

​Troubleshooting

​401 Unauthorized

​402 Payment Required - Insufficient Credits

​404 Not Found

​422 Validation Error

​429 Too Many Requests

​Invalid model string format

​Provider temporarily unavailable

​Next Steps

Frequently Asked Questions

General

What is Eden AI V3?

Which endpoint should I use?

How does the model string format work?

Authentication & Access

How do I get an API key?

Can I use my own provider API keys?

What are the rate limits?

How much does it cost?

File Handling

What file input methods are supported?

How long are uploaded files stored?

What are the file size limits?

Which file formats are supported?

LLM Features

Why is streaming mandatory in V3?

How do I send images to LLMs?

Which providers support vision/multimodal?

Does V3 support tool/function calling?

Universal AI

How do I discover available features?

Can I use fallback providers?

How do I optimize costs?

I’ve been an Eden AI user for some time. Is the previous version going to disappear?

Troubleshooting

401 Unauthorized

402 Payment Required - Insufficient Credits

404 Not Found

422 Validation Error

429 Too Many Requests

Invalid model string format

Provider temporarily unavailable

Next Steps