[FEATURE]: Make rate limit handling configurable with custom request limits #26314

Feature hasn't been suggested before.

• I have verified this feature I'm about to request hasn't been suggested before.

Describe the enhancement you want to request

Problem

When using OpenCode with certain LLM providers (for example, services with stricter rate limits like NVIDIA NIM), I quickly hit 429 rate limit errors. It seems the current implementation uses a fixed or inflexible limit (around 40 requests), which doesn't work well across different backends that enforce lower concurrency or RPM limits.

This leads to frequent failures and poor user experience on providers with more conservative quotas.

Proposed Solution

Add support for a user-configurable request limit / concurrency setting. This would allow setting any custom number (e.g., 10, 15, 20, etc.) instead of being restricted to the current fixed value.

Ideally this could be set via:

• Command line flag/argument

• Environment variable

• Project-level config file (if applicable)

This small addition would greatly improve compatibility with a wide range of LLM providers and deployment setups.

Additional Context

• The feature should be optional with a sensible default for users who don't need to tweak it.

• It would help both local/self-hosted models and various cloud providers with differing rate limit policies.

Thanks for considering this! OpenCode is otherwise working great.