Announcing $9M in funding led by NEA. Read more →
Higher performance and lower cost than any single LLM.
We invented the first LLM router. By dynamically routing between multiple models, Martian can beat GPT-4 on performance, reduce costs by 20%-97%, and simplify the process of using AI.
Switching takes just a few lines...
1import openai
2openai.api_key = "your_openai_api_key"
3
4chat_completion = openai.ChatCompletion.create(
5 model="gpt-3.5-turbo",
6 messages=[{"role": "user", "content": "Hello world!"}]
7)
8print(chat_completion.choices[0].message.content)Rocket performance
By using the best performing model for each request, we can achieve higher performance than any single model.
Martian outperforms GPT-4 across OpenAI's own evals (openai/evals).
Powered by Model Mapping.
A new interpretability Framework
Turn opaque black-boxes into interpretable representations.
Our router is the first tool built on top of our "Model Mapping" method. We are developing many other applications under this framework, including turning transformers from indecipherable matrices into human-readable programs.
Flying high on uptime
Get kicked off an API? We automatically find you the best alternative.
If a company experiences an outage or high latency period, automatically reroute to other providers so your customers never experience any issues.
Reduce your AI costs up to 98%
Don't waste money by paying senior models to do junior work. The model router sends your tasks to the right model.
Improve your product performance
Ensure you are always using the best model without the need to spend dozens of hours on engineers testing these models directly.
Install in seconds
The Martian API is dead simple to use. Import our package. Add your API key. Change one line of code where you're calling your LLM.
Boost your uptime
If a company experiences an outage or high latency period, automatically reroute to other providers so your customers never experience any issues.
Cost Calculator
Alien
efficiency
Determine how much you could save by using the Martian Model Router with our interactive cost calculator.
Input your number of users, tokens per session, sessions per month, and specify your cost/quality tradeoff
For the last 2.5 years, we've conducted research on evaluating and optimizing the performance of large language models. We've developed a method for predicting the performance of a model without running it.
This makes us the only people who can route to the best model without having to run all of the other models first.
Demo the router yourself
Match the optimal AI model for any prompt
