Add prefix for AI client model, to make it more flexible. Instead of relying on gpt prefix of gpt-4 or claude preifx of claude-haiku add optional provider prefix. E.g. openai::o1-mini, anthropic::claude-haiku, openai::new-weird-model-name.
This simple change will allow to reuse completion functions for any bullshit model name those companies will come up
Allow passing raw JSON request
Each provider has unique structure with interesting options. Allow passing those JSON options into request and apply them for request. This will allow to create bespoke requests. E.g. hermes chat --raw-options '{"config": "some config"}' -> raw options included in request
Extend provider selection
Add prefix for AI client model, to make it more flexible. Instead of relying on
gptprefix ofgpt-4orclaudepreifx ofclaude-haikuadd optional provider prefix. E.g.openai::o1-mini,anthropic::claude-haiku,openai::new-weird-model-name. This simple change will allow to reuse completion functions for any bullshit model name those companies will come upAllow passing raw JSON request
Each provider has unique structure with interesting options. Allow passing those JSON options into request and apply them for request. This will allow to create bespoke requests. E.g.
hermes chat --raw-options '{"config": "some config"}'-> raw options included in request