Tavily

“I like Tavily, easy to implement and super accurate retrieval. Others are ok but Tavily is just stronger imo”

“OpenCode's built-in websearch and webfetch work, but they're basic. If your agent hits a Cloudflare-protected page, it gets blocked. If it needs real search results, you're either paying for Tavily/Serper/Firecrawl or glueing together SearXNG + a headless browser + an extractor. I built Hound as a drop-in MCP server that handles all of that. No API keys, no config file, no paid tier. Just add it to your opencode.json and disable the built-in web tools. UPDATE!!!!: I AM LOST, I AM LOCKED OUT OF M”

“Don't sleep on this. Tavily Search has been a game changer for my team. You really don't know what you are missing until you try. I've yet to find a built in websearch that is more effective at finding technical facts. Highly recommended. Not to mention, we've embedded it as a core service in our Code Execution as a Service offering. Beta access available now. https://app.apifunnel.ai/”

“Ta∨іⅼу is probably the most agent-friendly option right now, specifically because it returns structured results designed for LLM consumption rather than raw HTML. SerpAPl works too but you're paying for a lot of overhead you don't need. LLMLayer is worth considering if you want search plus scraping in one API call. The model-agnostic angle matters if you're not locked into OpenAl's ecosystem.”

“Disclosure: I built this. Not selling anything here : the tier I'm describing is free, unlimited, and has no card behind it. I run a web search API for agents (SERPdive). It returns cleaned page content instead of links, so your model gets the text and not a list of URLs to go fetch itself (like Tavily or Exa) I just added a free model, Krill. Free and unlimited (fair use) : one request at a time, low priority. What you get per search: - ~700 tokens, vs ~1700 for Tavily to send to your llm - usu”

“Guys, we are actively scaling our research agent right now. We've been using Tavily for the last six months - great tool, no doubt, outta the box LangChain integration, but we are hitting tark now, and the API bill is starting to hurt unit economics ($15 for 1000 requests on advanced search is a tad too much) Google Custom Search is drastically changing their terms right now. Tested Exa.ai, but on our benchmarks for multilingual queries (non-English) the results weren't quite what we expected Fo”

“I am not looking for self-host but cheapest and best value out there in term of web search as a tool for agents. I am open to any framework as well. I know OpenAI has Tivily and others but I run into the free limit very fast. I need a bit higher limits lolz Same with Azure AI Foundry which is $$ after awhile. Perplexity Pro is same, I run into its monthly credit limit too. Any recommendation?”

“Honestly the search API costs add up way faster than you'd expect, especially when you're iterating on agent workflows. I've been down this exact rabbit hole and found a few things that actually work without breaking the bank. First, SerpAPI has pretty reasonable pricing if you're doing bulk searches, way better than Tavily's per-query costs once you hit volume. But here's the thing most people miss - you don't always need a search API at all. For a lot of use cases, we ended up building direct ”