AWS guide pairs OpenCode with Kimi K3, GPT-OSS 120B and Nemotron 3 Super as a pay-per-use open-weight coding agent
AWS Machine Learning Blog·low signal
AWS published an opencode.json setup that sends plan-mode work to Kimi K3 (1M context), uses GPT-OSS 120B as the default, and sends build-mode work to Nemotron 3 Super 120B for throughput. Latency-tolerant batch refactors can run on the Bedrock Flex tier at 50% off, and global cross-Region inference costs about 10% less. It is a concrete recipe for splitting planning and execution across cheaper open models inside a terminal agent.
AWS Machine Learning Bloglow signal