PHSa meron nang You do not have permission to view the full content of this post.
Log in or register now. experience, ano ang mga useful features niya para sa inyo?
Planning to deploy it on a small VPS.
Planning to deploy it on a small VPS.
-
Welcome to PHCorner Forums! Sign up for free to unlock full access and exclusive privileges, including:
- Asking questions about computer and mobile issues
- Discovering internet tips for your mobile network and computer
❓ Help May OpenClaw user ba dito?
- Thread starter Draft
- Start date
Skia Monarch
Elite
Gusto ko sana praktisan kaso iniisip ko pa use case kung gagawa ako gusto ko sana ilagay sa cloud para 24/7 wala na kasi ako extra storage sa pc kasi uunahin ko padin yung games ko na i store dito haha di pa invested. chatbot lang muna ako for now perplexity
Pang-cli and terminal apps lang yan like openclaw, claude code, opencode.... Kung gusto mo ng auto switch model to save costs, yung flow-llm-router sa github ang gamitin mo tapos i-chain mo sa litellm for safety and optimization. Pero yung rtk, designed talaga to reduce token usage (only) kaya prone to miss data sa malaking trabaho. Matipid, mapili sa apps at ang hirap i-setup. Yung flow-llm-router+litellm mas madali at ang focus nyan ay to save cost by switching from one model to another depende sa gravity ng tasks at may option for fallback models kung bumigay yung isa for continuous use. Parang similar sa openclaw.json config kung solo siya. Kailangan mo lang ng maraming live model apis kahit mga free ones na nakakalat sa net he he.
parang sa openclaw nga. Setting ko sa openclaw is fallback sa ollama kapag lite task lang since naka self hosted yung ollama ko tapos main agent ko is GPT4.1 from github copilot since free yung token cost from my $10 per month sa github copilot for software development. Since lite lang use cases ng openclaw namin for docs, enough na ang GPT4.1
Namali ako kanina, kaya edited ko yung last post. Yung flow-llm-router yon hindi free-llm-router. Pero kung mag-search ka sa GH, nandoon yung mga free AI api with router na he he.
You do not have permission to view the full content of this post. Log in or register now.
Subukan mong pandagdag yung ds-free-api sa GH. Unli yan, using 1 email
+ as many as you could add na same email sub-accounts (check mo sa issues section sa GH yung tips) with same password sa .env file. Two models yon: deepseek-default, deepseek-expert (latest ds4).
Ganito itsura pag-andar. Tested ko kaninang madaling-araw.
Free sa DS yung web platform nila, walang subscription di tulad sa kanilang api use. Di ko alam yung limits pero mag-vary ng 10 -30 rpm. Free sa public yan kung tutuusin, so create more sub-accounts of you email, he he.
He he, strike while the iron is hot. Most of the time, madali yan mawala dahil "nsfw" yan sa larangan ng AI din. Workaround siya kumbaga. Kung sakaling sumagot ng Chinese, pwersahin niyo sa system prompt or just tell it to answer in English or Tagalog. Na-stress test ko yung expert model with web search; kinaya nyang i-setup yung yung app na kakagawa lang by just mentioning the name and supplied a 2nd alternative code to setup automatically on its own na working din. Pwede na!
Translate mo sa browser. Lagyan mo ng alias yung default at expert model sa settings then test para yon yung magiging model id nya. Huwag mong kalimutan i-save na nasa baba ng page.
Spoiler contents are visible only to Established Members.
May free models din sa OpenRouter, ito mostly ang gamit ko.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
Yan yung api gateway ko noon. Kung maubos yung openrouter daily limit, mag-switch siya sa nilagay kong provider api keys sa BYOK section nya. Kahit Ollama supported niya rin using the ollama cloud models. Sabayan mo ng free Groq api, Cerebras. Gemini, Nvidia, Mistal atbp. matagalan yung gamit. At pwede yang tumanggap ng maraming apis of the same provider, parang Big-AGI feature.
Dagdag ko lang ito. Pag naglagay ka ng api mo sa mga online gateways, may tendency na manakaw yon, tulad ng Gemini apis ko atbp. na laging ubos kahit di ko ginagamit. Mas mainam may sarili kang ai gateway or llm router/manager. controlled mo pa at madaling i-monitor yung usage.
Subukan mo yung dalawang router sa last spoiler ko. Kasama na openrouter dyan atbp in one go. Pwede ka rin magdagdag ng anthropic o openai compatible api na wala sa list. Yan yung gold standard ngayon sa llm community.
Multi-accounts ng groq.com? Pwede nga yan dahil galante siya sa medium open source models. Ang naisip kong free method, gagamit ng tipong multi-proxy app (psiphon w/ quic servers, v2ray na vless+reality) para may 70-80% chance na di ka ma-ban, temp mails, anti-detect browser na say with 5 profiles ng 5 apis dyan each using the proxy locations to create the apis, llm router gateway na tumatanggap ng multi-apis - halos unli na ~You do not have permission to view the full content of this post. Log in or register now.. Kalahati sa free model list ay patay, kaya verify muna individually bago gamitin.
Sorry kung namali ako ng basa noong una "groq" pala sinabi mo - my bad. Grok2api still works, pero grok.com models yon, hindi yung nasa "groq" api site nila na open source models. Pandagdag din yan for small burst of real time research. Mahigpit yung rate limiting from 5 req/24 hrs to ~10 req/2 hrs depende sa model. Tiyagaan na lang sa reset time.
Similar threads
-
-
👨🏫 Tutorial LibreChat AI - How To Host Your Own AI
- Started by Soul Calibre
- Replies: 145
-
🕯️ Traditions Bumalik ako dito galing 2066. May tanong ba kayo?
- Started by PQ4
- Replies: 450