Mike, Nathan, and PickAxe Team -
I’m curious as to why we’re not permitted to utilize free options that could increase our profits when we charge users, such as the OpenRouter free models (which offer about 15-20) or the free GPT-4o and 4o-Mini keys provided by GitHub. Despite token throttling, this could be a viable option for deployed studios. Perhaps I’m overlooking something?
Thanks,
Jason
I really wanna see the Gemini and Llama models. Gemini is so good at long context retrieval plus the token cost for Flash is very cheap and its multimodal too. And if something like llama and groq are supported, they can be combined with elevenlabs action. That way, we might be able to have very low latency audio bots on pickaxe.
I think openrouter might be too complex for the team to work consistently with pickaxe actions reliably. That’s probably why the selection is limited.