Supported Providers

Complete list of 78+ AI providers supported by Portkey AI Gateway

Overview

Portkey AI Gateway supports 250+ LLMs from 78+ providers, giving you access to virtually every major AI model through a single, unified API.

All Supported Providers

Here’s the complete list of providers integrated with Portkey:

Major LLM Providers

Provider	Models	Features	Documentation
OpenAI	GPT-4, GPT-3.5, o1, o3, DALL-E, Whisper	Chat, Completions, Embeddings, Images, Audio, Realtime	OpenAI →
Anthropic	Claude 3.5, Claude 3, Claude 2	Chat, Vision, Function Calling	Anthropic →
Azure OpenAI	GPT-4, GPT-3.5, Embeddings	All OpenAI features via Azure	Azure OpenAI →
Google Gemini	Gemini 2.0, Gemini 1.5 Pro/Flash	Chat, Vision, Embeddings, Function Calling	Google Gemini →
AWS Bedrock	Claude, Llama, Mistral, Titan	Chat, Embeddings, Converse API	AWS Bedrock →
Cohere	Command, Command R, Command R+	Chat, Embeddings, Rerank	Cohere →
Mistral AI	Mistral Large, Medium, Small	Chat, Embeddings, Function Calling	Mistral →

Specialized Providers

Provider	Models	Specialty	Documentation
Together AI	100+ open models	Open-source models, Fast inference	Together AI →
Anyscale	Llama, Mistral, Mixtral	Open models with Endpoints	Anyscale →
Groq	Llama, Mixtral, Gemma	Ultra-fast inference (500+ tokens/s)	Groq →
DeepInfra	100+ models	Cost-effective inference	DeepInfra →
Perplexity	Sonar models	Search-augmented generation	Perplexity →
Ollama	Any local model	Local/self-hosted models	Ollama →
Fireworks AI	80+ models	Fast inference, fine-tuning	Fireworks AI
Replicate	Thousands of models	Community models, image gen	Replicate

Cloud AI Platforms

Provider	Description
Google Vertex AI	Google Cloud AI platform with Gemini, PaLM
Azure AI Inference	Microsoft’s unified AI inference service
Sagemaker	AWS machine learning platform
Workers AI	Cloudflare’s edge AI platform

Additional Providers (A-Z)

A-C Providers

• 302.AI - AI model aggregation platform
• AI21 - Jamba models
• AIBadgr - Educational AI platform
• Anyscale - Ray-based inference
• Cerebras - Ultra-fast inference
• CometAPI - API marketplace
• Cohere - Enterprise NLP
• Cortex - Snowflake AI

D-I Providers

• DashScope - Alibaba AI platform
• DeepBricks - AI infrastructure
• DeepInfra - Cost-effective inference
• DeepSeek - Chinese AI models
• Featherless AI - Lightweight models
• Fireworks AI - Fast inference platform
• HuggingFace - 100,000+ models
• Hyperbolic - Decentralized AI
• Inference.net - Distributed inference
• IO Intelligence - Enterprise AI

J-O Providers

• Jina - Embeddings and search
• Kluster AI - Cluster computing
• Krutrim - Indian AI models
• Lambda - GPU cloud
• LemonfoxAI - AI infrastructure
• Lepton - Simplified AI deployment
• LingYi - Chinese AI models
• MatterAI - Scientific AI
• Meshy - 3D generation
• Milvus - Vector database
• Modal - Serverless AI
• MonsterAPI - Cost-effective inference
• Moonshot - Chinese AI platform
• NCompass - Enterprise AI
• Nebius - Cloud AI platform
• NextBit - AI infrastructure
• Nomic - Embeddings (Nomic Embed)
• Novita AI - Multi-modal AI
• NScale - Scalable inference
• Ollama - Local models
• OpenRouter - Model router
• Oracle - Oracle Cloud AI
• OVHcloud - European cloud AI

P-Z Providers

• PaLM - Google’s legacy models
• Perplexity AI - Search-augmented LLMs
• Predibase - Fine-tuning platform
• Qdrant - Vector search
• Recraft AI - Image generation
• Reka AI - Multimodal models
• Replicate - Community model hosting
• SambaNova - AI hardware acceleration
• Segmind - Image generation
• SiliconFlow - Chinese AI platform
• Stability AI - Stable Diffusion
• Together AI - Open-source models
• Triton - NVIDIA Triton
• Tripo3D - 3D generation
• Upstage - Korean AI models
• Voyage - Embeddings
• Workers AI - Cloudflare edge AI
• X.AI - Grok models
• Z.AI - AI infrastructure
• Zhipu - Chinese AI (ChatGLM)

Provider Identifier Reference

When making requests, use these provider identifiers:

# Syntax
client = Portkey(
    provider="<provider-identifier>",
    Authorization="<api-key>"
)

Common Provider Identifiers

Provider Name	Identifier	Example
OpenAI	openai	provider="openai"
Anthropic	anthropic	provider="anthropic"
Azure OpenAI	azure-openai	provider="azure-openai"
Google Gemini	google	provider="google"
AWS Bedrock	bedrock	provider="bedrock"
Cohere	cohere	provider="cohere"
Mistral AI	mistral-ai	provider="mistral-ai"
Together AI	together-ai	provider="together-ai"
Anyscale	anyscale	provider="anyscale"
Groq	groq	provider="groq"
Perplexity	perplexity-ai	provider="perplexity-ai"
DeepInfra	deepinfra	provider="deepinfra"
Ollama	ollama	provider="ollama"
Fireworks AI	fireworks-ai	provider="fireworks-ai"
Replicate	replicate	provider="replicate"

Feature Support Matrix

Core Features

Provider	Chat	Streaming	Embeddings	Function Calling	Vision
OpenAI	✅	✅	✅	✅	✅
Anthropic	✅	✅	❌	✅	✅
Azure OpenAI	✅	✅	✅	✅	✅
Google Gemini	✅	✅	✅	✅	✅
AWS Bedrock	✅	✅	✅	✅	✅
Cohere	✅	✅	✅	✅	❌
Mistral	✅	✅	✅	✅	❌
Together AI	✅	✅	✅	✅	✅
Anyscale	✅	✅	✅	✅	❌
Groq	✅	✅	❌	✅	✅
DeepInfra	✅	✅	❌	✅	✅
Perplexity	✅	✅	❌	❌	❌
Ollama	✅	✅	✅	❌	✅

Special Features

Provider	Audio (TTS)	Audio (STT)	Image Generation	Batch API	Fine-tuning
OpenAI	✅	✅	✅	✅	✅
Anthropic	❌	❌	❌	✅	❌
Azure OpenAI	✅	✅	✅	✅	✅
AWS Bedrock	❌	❌	✅	✅	✅
Stability AI	❌	❌	✅	❌	❌
Fireworks AI	❌	❌	✅	❌	✅

Request Examples

Basic Provider Switching

from portkey_ai import Portkey

# OpenAI
openai_client = Portkey(provider="openai", Authorization="sk-***")
response = openai_client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello"}]
)

# Switch to Anthropic - same code structure!
anthropic_client = Portkey(provider="anthropic", Authorization="sk-ant-***")
response = anthropic_client.chat.completions.create(
    model="claude-3-5-sonnet-20241022",
    messages=[{"role": "user", "content": "Hello"}]
)

Multi-Provider Fallback

config = {
    "strategy": {"mode": "fallback"},
    "targets": [
        {"provider": "openai", "api_key": "sk-***"},
        {"provider": "anthropic", "api_key": "sk-ant-***"},
        {"provider": "google", "api_key": "***"},
        {"provider": "groq", "api_key": "gsk-***"}
    ]
}

client = Portkey().with_options(config=config)

Adding New Providers

Portkey regularly adds new providers. To request a provider integration:

1. Check the GitHub issues for existing requests
2. Open a feature request with provider details
3. Contribute a provider implementation

Contribute a Provider

Help add new providers to the gateway

Provider Pricing

For detailed pricing information across all providers, visit:

Portkey Models

Browse pricing for 2,300+ models across 40+ providers

Next Steps

Provider Overview

Learn how provider routing works

OpenAI

OpenAI integration guide

Fallbacks

Set up automatic fallbacks

Load Balancing

Distribute across providers