Open Source Models for Check Run Agents
Choose from frontier and open source models to optimize cost, latency, and performance for every custom PR check.
Check Run Agents now support several new open source models:
kimi-k2-7-codeglm-5-2minimax-m3qwen-3-7-plus
Check Run Agents have become a popular way for teams to encode their own code review workflows into custom agents that run on every PR.
At the same time, the open source model ecosystem has reached an inflection point. There are now exceptional OSS models that are competitive with frontier models across many coding and reasoning tasks, each with their own tradeoffs around cost, latency, context window, and performance.
We think giving customers access to this ecosystem is especially important because many teams have built extensive suites of Check Run Agents that execute on every PR. At that scale, inference costs can become substantial.
If an open source model can deliver the same outcome for a particular agent as a frontier model, you should be able to make that tradeoff yourself.
Choose the model that's right for each agent. See our docs for the full list of supported models and pricing.
