Skip to main content

​
Why switch to CheapestInference?

  • Lower costs: Save up to 90% on inference costs
  • More models: Access to many open-source models
  • No vendor lock-in: Open-source models with transparent pricing
  • Better privacy: Your data stays private and secure

​
Migration guide

​
Using the official OpenAI SDK

You can use the official OpenAI Python or TypeScript SDKs with CheapestInference by simply changing the base URL and API key: 
from openai import OpenAI

# Just change these two lines!
client = OpenAI(
    api_key="your_cheapestinference_api_key",
    base_url="https://api.cheapestinference.ai/v1"
)

# Everything else works exactly the same
response = client.chat.completions.create(
    model="meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello!"}
    ]
)

print(response.choices[0].message.content)

​
Environment variables

For even easier migration, use environment variables: 
export OPENAI_API_KEY=your_cheapestinference_api_key
export OPENAI_BASE_URL=https://api.cheapestinference.ai/v1
Then your code doesn’t need to change at all: 
from openai import OpenAI

client = OpenAI()  # Automatically uses env vars

response = client.chat.completions.create(
    model="meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo",
    messages=[{"role": "user", "content": "Hello!"}]
)

​
Supported endpoints

All major OpenAI API endpoints are supported:
EndpointSupportNotes
/v1/chat/completions✅ FullIncluding streaming, function calling, and structured outputs
/v1/embeddings✅ FullMultiple embedding models available
/v1/models✅ FullList all available models
/v1/files✅ FullFile upload for fine-tuning
/v1/batches✅ FullBatch inference API

​
Model mapping

While CheapestInference offers many models, here’s how OpenAI models map to our alternatives:
OpenAI ModelCheapestInference AlternativeNotes
gpt-4meta-llama/Meta-Llama-3.1-70B-Instruct-TurboSimilar quality, 80% cheaper
gpt-4-turbometa-llama/Meta-Llama-3.1-405B-Instruct-TurboHigher quality, 70% cheaper
gpt-3.5-turbometa-llama/Meta-Llama-3.1-8B-Instruct-TurboFaster and cheaper
text-embedding-3-largeBAAI/bge-large-en-v1.5Best quality embeddings
text-embedding-3-smallBAAI/bge-small-en-v1.5Fast and efficient
whisper-1openai/whisper-large-v3Same model, lower cost
dall-e-3black-forest-labs/FLUX.1-schnellBetter quality, faster

​
Framework integrations

CheapestInference works seamlessly with popular frameworks:

​
LangChain

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo",
    openai_api_key="your_cheapestinference_api_key",
    openai_api_base="https://api.cheapestinference.ai/v1"
)

​
LlamaIndex

Python
from llama_index.llms.openai import OpenAI

llm = OpenAI(
    model="meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo",
    api_key="your_cheapestinference_api_key",
    api_base="https://api.cheapestinference.ai/v1"
)

​
Vercel AI SDK

TypeScript
import { openai } from '@ai-sdk/openai';

const provider = openai({
  apiKey: 'your_cheapestinference_api_key',
  baseURL: 'https://api.cheapestinference.ai/v1'
});

const model = provider('meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo');

​
Feature comparison

FeatureOpenAICheapestInference
Streaming✅✅
Function calling✅✅
Structured outputs✅✅
Embeddings✅✅
Batch API✅✅
JSON mode✅✅
Reproducible outputs✅✅

​
Differences to be aware of

While we strive for full compatibility, there are a few differences:
  1. Model names: Use our model naming format (e.g., meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo)
  2. Rate limits: Different rate limits apply (generally more generous)
  3. Model capabilities: Some models may have different context windows or capabilities
  4. Response times: Generally faster due to optimized infrastructure