Skip to main content

Overview

CyrionAI provides access to a wide range of state-of-the-art AI models across different modalities. All models are available at no cost for qualified nonprofit organizations.

Text Models

GPT Models

ModelDescriptionBest ForContext WindowCost
gpt-4o-miniFast and efficient GPT-4 modelGeneral tasks, quick responses128K tokensFree
gpt-4oMost capable GPT-4 modelComplex reasoning, creative writing128K tokensFree
gpt-3.5-turboLegacy modelSimple tasks, cost-effective16K tokensFree

Claude Models

ModelDescriptionBest ForContext WindowCost
claude-3-haikuFast and efficient Claude modelQuick responses, simple tasks200K tokensFree
claude-3-sonnetBalanced performance Claude modelGeneral purpose, good reasoning200K tokensFree
claude-3-opusMost capable Claude modelComplex analysis, creative writing200K tokensFree

Open Source Models

ModelDescriptionBest ForContext WindowCost
llama-3-8bMeta’s Llama 3 8B modelCustom applications, fine-tuning8K tokensFree
llama-3-70bMeta’s Llama 3 70B modelComplex reasoning, research8K tokensFree
mistral-7bMistral AI’s 7B modelGeneral purpose, efficient32K tokensFree

Image Models

DALL-E Models

ModelDescriptionBest ForResolutionCost
dall-e-2OpenAI’s DALL-E 2 modelHigh-quality realistic images1024x1024Free
dall-e-3OpenAI’s DALL-E 3 modelMost advanced image generation1024x1024Free

Stable Diffusion Models

ModelDescriptionBest ForResolutionCost
stable-diffusion-xlStability AI’s SDXL modelArtistic and creative images1024x1024Free
stable-diffusion-2.1Stable Diffusion 2.1General image generation768x768Free

Other Image Models

ModelDescriptionBest ForResolutionCost
midjourneyMidjourney modelHigh-quality artistic images1024x1024Free
fluxFlux modelFast image generation1024x1024Free

Video Models

ModelDescriptionBest ForDurationCost
pika-labsPika Labs modelHigh-quality video generation3-10 secondsFree
runway-mlRunway ML modelCinematic quality videos3-10 secondsFree
stable-video-diffusionStable Video DiffusionOpen-source video generation3-10 secondsFree

Audio Models

ModelDescriptionBest ForLanguagesCost
whisper-1OpenAI’s Whisper modelAudio transcription99+ languagesFree
whisper-large-v3Large Whisper modelHigh-accuracy transcription99+ languagesFree

Model Capabilities

Text Generation Features

  • Function Calling: Available on GPT-4 and Claude models
  • Web Search: Available on GPT-4 models
  • Image Understanding: Available on GPT-4 models
  • Multimodal Input: Support for text, images, and other media
  • Tool Integration: Ability to call external tools and APIs

Image Generation Features

  • Multiple Sizes: Support for various aspect ratios and resolutions
  • Quality Options: Standard and HD quality settings
  • Style Control: Ability to specify artistic styles and techniques
  • Variation Generation: Create multiple variations from a single prompt

Video Generation Features

  • Motion Control: Specify camera movements and actions
  • Style Transfer: Apply different visual styles to videos
  • Duration Control: Generate videos of varying lengths
  • Quality Options: Different quality settings available

Audio Transcription Features

  • Multilingual Support: 99+ languages supported
  • Language Detection: Automatic language identification
  • Timestamp Support: Get word-level timestamps
  • Format Options: JSON, text, SRT, and verbose JSON outputs

Model Selection Guide

For Text Generation

Choose GPT-4o-mini if:
  • You need fast responses
  • Working with simple tasks
  • Cost efficiency is important
Choose GPT-4o if:
  • You need the highest quality responses
  • Working on complex reasoning tasks
  • Creative writing is required
Choose Claude-3-sonnet if:
  • You need balanced performance
  • Working on analytical tasks
  • Want good reasoning capabilities

For Image Generation

Choose DALL-E 2 if:
  • You need realistic, high-quality images
  • Working on professional content
  • Want reliable, consistent results
Choose Stable Diffusion if:
  • You need artistic, creative images
  • Want more control over style
  • Working on experimental content
Choose Midjourney if:
  • You need highly artistic images
  • Working on creative projects
  • Want unique visual styles

For Video Generation

Choose Pika Labs if:
  • You need high-quality video generation
  • Want good motion and consistency
  • Working on professional content
Choose Runway ML if:
  • You need cinematic quality
  • Working on documentary-style content
  • Want professional video output

For Audio Transcription

Choose Whisper-1 if:
  • You need general transcription
  • Working with common languages
  • Want fast processing
Choose Whisper-large-v3 if:
  • You need high accuracy
  • Working with complex audio
  • Want better handling of accents and noise

Rate Limits

All models have generous rate limits for nonprofit organizations:
Model TypeRate LimitNotes
Text Models100 requests/minutePer API key
Image Models20 requests/minutePer API key
Video Models10 requests/minutePer API key
Audio Models50 requests/minutePer API key
Rate limits are per API key. Contact us if you need higher limits for your specific use case.

Model Updates

We regularly update our model offerings to provide access to the latest AI capabilities. New models are added automatically to your account at no additional cost.

Recent Updates

  • GPT-4o-mini: Added for faster, more efficient text generation
  • Claude-3 models: Added for enhanced reasoning capabilities
  • DALL-E 3: Added for improved image generation quality
  • Video models: Added for video generation capabilities

Getting Started

To use any of these models, simply specify the model name in your API requests:
import openai

client = openai.OpenAI(
    api_key="your-api-key",
    base_url="https://ai.cyrionlabs.org/v1"
)

# Text generation
response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello!"}]
)

# Image generation
response = client.images.generate(
    model="dall-e-2",
    prompt="A beautiful sunset"
)

# Video generation
response = client.videos.generate(
    model="pika-labs",
    prompt="A peaceful landscape"
)

# Audio transcription
with open("audio.mp3", "rb") as audio_file:
    response = client.audio.transcriptions.create(
        model="whisper-1",
        file=audio_file
    )

Model Comparison

For detailed comparisons between models, see our Model Comparison Guide.

Need Help?

If you’re unsure which model to use for your specific use case: