This website uses cookies to anonymously analyze website traffic using Google Analytics.
Company

Together AI acquires CodeSandbox to launch first-of-its-kind code interpreter for generative AI

December 12, 2024

By 

Together AI

Code generation has quickly become one of the most popular use cases for LLMs. However, while LLMs excel at generating code, they cannot execute it, leaving developers to manually test, debug, and implement the generated code in separate environments. This limitation slows down workflows and prevents LLMs from addressing real-world challenges that require immediate, actionable outputs.

That is changing today. We’re thrilled to announce that Together AI has acquired CodeSandbox, and we are joining forces to make the Together Inference Platform the first of its kind to feature built-in code interpretation through the Together Code Interpreter.

Launched in 2017 and based in Amsterdam, The Netherlands, CodeSandbox pioneered a unique development environment infrastructure used by more than 4.5 million developers every month. CodeSandbox enables developers to spin up virtual machine (VM) sandboxes for code execution, hibernate them, and resume nearly instantly—offering unparalleled performance, security and scale. We are thrilled to join forces with CodeSandbox and begin a new chapter of generative AI.

Unleashing the full potential of LLMs

Imagine allowing a state-of-the-art LLM like Llama 3.3 70B to process and visualize real-time sales data, simulate an algorithm for route optimization, or test and debug application code in an actual development environment. This is exactly what we are doing by bringing a code interpreter natively to the Together API.

With Together Code Interpreter, we allow our shared community of millions of developers to harness the full potential of LLMs by giving them the ability to execute code. This technology takes LLM-generated code as input, executes it in a fast, isolated, and secure environment, and feeds the output back to the LLM. As a result, it gives any LLM the ability to seamlessly perform data analysis, visualize data, and solve more complex problems—allowing developers to build impressive gen AI applications.

The Together Code Interpreter has already generated excitement among developers at several AI-native companies, including Blackbox AI.

“More than 10 million monthly users rely on blackbox.ai’s coding agents to transform how they build software today. Thanks to CodeSandbox SDK, these agents execute the generated code securely,” said Blackbox AI Co-Founder and COO Robert Rizk. “We found this technology to be the best on the market today to sustain a very high load of requests and appreciate the unique support from the team behind it.”

{{custom-cta-1}}

To extend this breakthrough to every developer, today we are also thrilled to share that CodeSandbox is launching CodeSandbox SDK beta: an API to create and run VM sandboxes quickly and securely to run untrusted code at scale.

For those wanting to start experimenting with this technology before our Together Code Interpreter public rollout, we have published a quickstart guide on how to use CodeSandbox SDK to bring code interpretation to your apps calling Together serverless endpoints.

"At Together AI, we provide a platform that empowers developers and businesses to manage the entire generative AI lifecycle with unmatched performance, control, and cost-efficiency,” said Together AI Founder and CEO Vipul Ved Prakash. “As LLMs evolve to solve increasingly complex programming problems, integrating CodeSandbox’s code interpretation technology into our platform will enable developers to build powerful applications that seamlessly execute complex logic and invoke diverse tools to address real-world challenges. This marks a significant milestone for making generative AI more accessible, versatile, and impactful for our growing community."

Building powerful developer tooling

This is just the beginning of what the Together Code Interpreter makes possible. Thanks to this technology, we will allow LLMs to spin up browsers, search the web, draw charts, run evals, and further enhance how companies set up complex agentic workflows.

You can see this in action below in a teaser of the new version of LlamaCoder we are going to launch next week, which will run CodeSandbox SDK under the hood to execute the code of the applications it writes using Llama 3.1 405B.

A new era for open-source GenAI development

Together Code Interpreter marks a pivotal moment for developers and businesses working with generative AI. By bringing seamless code interpretation to our platform, we are breaking down barriers that have long limited the potential of LLMs, empowering developers to create applications that are capable of solving real-world problems with efficiency and precision.

This acquisition of CodeSandbox embodies our commitment to open-source AI and our vision of a future where every developer has the tools to innovate without limitations. With the Together Code Interpreter and CodeSandbox SDK, we’re not just enhancing developer workflows—we’re redefining how GenAI applications are built, deployed, and scaled.

We’re excited to see how our growing community of developers leverages this technology to build the next generation of AI-powered applications. Together, we’re shaping the future of generative AI and ensuring that innovation remains open, transparent, and within reach for everyone.

  • Lower
    Cost
    20%
  • faster
    training
    4x
  • network
    compression
    117x

Get early access to Together Code Interpreter

Bring complex logic, code execution, data analysis, and more to your GenAI application.

Q: Should I use the RedPajama-V2 Dataset out of the box?

RedPajama-V2 is conceptualized as a pool of data that serves as a foundation for creating high quality datasets. The dataset is thus not intended to be used out of the box and, depending on the application, data should be filtered out using the quality signals that accompany the data. With this dataset, we take the view that the optimal filtering of data is dependent on the intended use. Our goal is to provide all the signals and tooling that enables this.

Sign up for early access

Join our waitlist to be one of the first to use the Together Code Interpreter and bring the power of code interpretation to your GenAI application.

No items found.
Start
building
yours
here →