Costs is too high

[TidalWave123]

I did 4 job searches, 1 resume created and an update to the code via claude. That cost me $20 USD in tokens.

[MoritzRodermund]

I used Kimi K2.6 through Ollama & OpenCode and immensly reduced costs… can recommend trying it out…

[tho-maj]

you run kimi k2.6 locally? :D

[TidalWave123]

any instructions on how to re-write the code so i can use the local LLM instead of the high cost. I am a bit lost on how to enable the LLM to take over the majority of the steps. I can still use claude tokens if needed just trying to lower the majority of the cost.

[umrashrf]

@TidalWave123 I could be wrong but he is using claude cli which is kinda vendor locked in and it's closed source. gemini is better and has free tier. But it should be definitely use ollama local first agentic cli.

[umrashrf]

Cost is the reason for me that's why I stick to gpt-4o model. Is it possible to use gpt-4o or local gemma4 or qwen3?