Replit, the collaborative software development and deployment platform, has announced the launch of ModelFarm, a new offering designed to provide developers easy and secure access to generative AI models. With ModelFarm, Replit aims to accelerate the development of AI applications by eliminating the hassles of API key management.
ModelFarm provides a centralized portal for developers to integrate leading generative AI models into their applications without worrying about API credentials leaking or exceeding usage quotas. Replit handles the authentication and usage tracking in the background, allowing developers to focus on building.
from replit.ai.modelfarm import CompletionModel
model = CompletionModel("text-bison")
prompts = ["What is the best way to learn AI?"]
completion = model.complete(prompts, max_output_tokens=50, temperature=0.2)
print(completion.responses[0].choices[0].content)Sample code to create a simple app
Replit says subscribers to its Hacker and Pro plans will get free access to the following models from Google Cloud Vertex AI until October 15:
- Chat model:
chat-bison - Text Completion:
text-bison - Text Embedding:
textembedding-gecko
Support for integrated billing and monitoring is planned for the future, enabling developers to set usage quotas and alert thresholds as their applications scale up.
By eliminating the overhead of API key management, ModelFarm represents a big step towards Replit's vision of empowering every software builder to become an AI developer. The company hopes ModelFarm will unlock new waves of creativity as developers integrate generative AI into their applications with ease.
The announcement comes on the heels of Replit's recently launched Autoscaling and Static Deployments features, underscoring their focus on enabling frictionless development of production-ready apps. With innovations like ModelFarm, Replit aims to cement its position as the go-to cloud development platform for the AI era.