--- title: If you don't run your own LLM proxy, how do you know... description: A developer argues that relying on AI providers' built-in usage dashboards is insufficient for real engineering decisions, since those panels lack time-series... canonical: https://daily.dev/posts/if-you-don-t-run-your-own-llm-proxy-how-do-you-know-whats-going-on--3sryq76zv twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: If you don't run your own LLM proxy, how do you know whats going on? | daily.dev og:description: A developer argues that relying on AI providers' built-in usage dashboards is insufficient for real engineering decisions, since those panels lack time-series... og:url: https://daily.dev/posts/if-you-don-t-run-your-own-llm-proxy-how-do-you-know-whats-going-on--3sryq76zv og:image: https://api.daily.dev/og/posts/3srYQ76ZV.png og:image:alt: If you don't run your own LLM proxy, how do you know whats going on? og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # If you don't run your own LLM proxy, how do you know whats going on? **[Hacker News](https://daily.dev/sources/hn)** · 3 min read · 0 upvotes · 0 comments ## Summary A developer argues that relying on AI providers' built-in usage dashboards is insufficient for real engineering decisions, since those panels lack time-series token data and meaningful observability. The piece advocates running your own self-hosted LLM proxy to gain multi-provider/multi-model routing through a single endpoint, detailed logs of which key called what model and when, virtual API keys to avoid scattering a single main key across tools and agents, full control over infrastructure, and the ability to quickly switch models or providers without redoing setup. The author describes modifying their own proxy for better insight and custom MCP endpoints, and notes they can diagnose cache misses in about 20 seconds using the data it provides. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why would I want to run my own LLM proxy instead of using the provider's built-in analytics dashboard? Provider analytics panels typically show surface-level usage stats, not engineer-grade time-series token data. A self-hosted proxy gives you a single endpoint across multiple providers and models, logs of which API key called what model and when, virtual API keys so you don't scatter one main key across agents and tools, full infrastructure control, and the ability to switch providers quickly without redoing setup. _Developers weighing proxy setups for LLM observability can compare approaches and tooling discussions on daily.dev._ ### What are the benefits of using virtual API keys with an LLM proxy? Virtual API keys let you avoid littering a single main provider key across your work, agents, and harnesses. By issuing separate virtual keys per use case through a self-hosted proxy, you gain per-key logs of which client called which model and when, along with better security isolation and easier auditing than sharing one key everywhere. _Teams managing multiple AI agents and tools can track key-management best practices via daily.dev._ ## Community take How the wider developer community reacted, aggregated from 2 discussions and 97 comments across hackernews (as of 2026-10-19). **TL;DR:** Discussion mostly veers away from the proxy argument itself into whether API pricing versus subscriptions makes sense, with many sharing proxy/gateway tools (LiteLLM, GoModel, Bifrost, Aperture) while others question whether this level of monitoring is worth the effort for typical usage. **Sentiment:** 30% positive · 45% mixed · 25% skeptical **The case for** - Self-hosted proxies/gateways give virtual keys, detailed per-key/per-model logs, and guardrails like prompt-injection checks. - Several people vouch for specific tools (GoModel, Bifrost, Aperture, agentsview) as solid, actively maintained options. - Tracking usage closely helps avoid nasty surprises if subsidized subscription pricing eventually gets replaced by real API costs. - Running a mix of open-weight and frontier models keeps you from becoming overly locked into one provider. **The pushback** - Some argue this is over-engineering/setup-theatre for a problem a bigger plan or simpler prompt tuning would solve. - Concerns that LiteLLM and similar proxy tools are buggy, fast-changing, and possibly insecure 'vibecoded' projects. - Skepticism that typical hobby usage ($20-100/month) ever approaches the scale where this kind of observability matters. - Using managed harnesses like Codex while claiming to 'own your infrastructure' is seen as contradictory. **By community** - hackernews (mixed): Conversation splits between people sharing self-hosted proxy/gateway setups they like and others questioning whether the effort is justified, with a long tangential debate about API vs subscription pricing economics. **Hottest debate:** Whether AI subscription pricing is a sustainable deal or a loss-leader 'drug dealer' tactic that will eventually force users onto expensive API pricing. **Open questions** - Is there a reliable way to archive and search agent/conversation logs across multiple harnesses (Claude, Codex, etc.)? - Will AI providers' subsidized subscription pricing hold up long-term, and how would that affect proxy/observability strategies? - What concrete product or business decisions are actually being made from this level of token-level observability, beyond cost tracking? **Highlights** > The current crop of flash models are more comparable to sonnet than opus but they are very affordable. The reality is that the subscriptions are heavily subsidized. Pretty much every sub that's not run by one of the money-burning frontier labs has had to drop it or dramatically increase prices and lower limits in the last few months. Cheap tokens are an acquisition tool, not a real product. It sounds dramatic but it really is the drug dealer sales funnel. At some point they're going to drop the real prices in your lab and you're going to have to figure out how to pay that much or pay someone else a lot less. Better to get a handle on it now. > — [dcss\_gardener on hackernews · 5 comments](https://news.ycombinator.com/item?id=50147398) > I agree its so important to proxy and capture all the data. I've been very happy with bifrost as a proxy. It's a quality piece of software, unlike LiteLLM. You also get this one interface to access all your models, across all services, including self-hosted, STT models, everything. There are ways to use subscriptions with proxies too, but they are discouraged because of grey-hat uses.. > — [zaptheimpaler on hackernews · 1 comments](https://news.ycombinator.com/item?id=50148594) > I use GoModel and its amazing. My favorite features are virtual model, the support for arbitrary request labels, and the nice dashboads / ui. Creator has more sophisticated report generation on the near term roadmap. He's super responsive to feedback and ships fast. (except for his recent trip to the US west coast to meet founders and investors, he's EU based) You can use and control all.your subscriptions or aoikeys through a gateway like this, so you don't get pwned like what just happened with OpenCode (or an OpenRouter like vendor). GoModel has some ambitions like this, but is open source so you can run it on you own infra, with plans for self hosted enterprise as well (there are a few features already like SSO). We are rolling out the free version at work right now. I run GoModel at home and have hyper detailed history of every agent interaction, across models running on my sparks and OpenCode sub. You can call me a committed superfan of this project. https://github.com/ENTERPILOT/GOModel > — [verdverm on hackernews](https://news.ycombinator.com/item?id=50147194) > All that posturing for the purpose to be manually monitoring cache hits and misses. This can't be serious. Buy a larger plan or reduce some instructions and you don't have to deal with this waste of time. > — [vasco on hackernews](https://news.ycombinator.com/item?id=50149197) > > Own your infrastructure ...proceeds to use codex. With just a 12gb VRAM GPU + 64gb of CPU RAM, and you can actually own your own infrastructure. I can't wait for hardware prices to fall so that becomes even more accessible. > — [kennywinker on hackernews](https://news.ycombinator.com/item?id=50147538) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=50145507) · 150 points · 97 comments - [hackernews](https://news.ycombinator.com/item?id=50138678) · 4 points · 0 comments ## Similar posts on daily.dev - [The True Cost of Unmanaged LLM Usage in Enterprise](https://daily.dev/posts/the-true-cost-of-unmanaged-llm-usage-in-enterprise-9a5dilhuj) · Medium · 0 upvotes · 0 comments - [Multi-Provider LLM Routing Is Not a Problem, It's Your Architecture: Inference in Production Series](https://daily.dev/posts/multi-provider-llm-routing-is-not-a-problem-it-s-your-architecture-inference-in-production-series-b8ljpdbw8) · DigitalOcean Community · 0 upvotes · 0 comments - [Untitled](https://daily.dev/posts/untitled-m9iztiroa) · SitePoint · 0 upvotes · 0 comments - [My self-hosted LLMs are a lot more than just a chat replacement – here's how they boost my productivity](https://daily.dev/posts/my-self-hosted-llms-are-a-lot-more-than-just-a-chat-replacement-here-s-how-they-boost-my-productiv-dqzrp8yzq) · XDA Developers · 0 upvotes · 0 comments - [How LLM Routing Actually Works in Production](https://daily.dev/posts/how-llm-routing-actually-works-in-production-rsf5hw9mf) · Daily Dose of Data Science \| Avi Chawla \| Substack · 2 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#observability](https://daily.dev/tags/observability), [#self-hosting](https://daily.dev/tags/self-hosting) [View this post on daily.dev](https://daily.dev/posts/if-you-don-t-run-your-own-llm-proxy-how-do-you-know-whats-going-on--3sryq76zv) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"If you don't run your own LLM proxy, how do you know whats going on?","url":"https://daily.dev/posts/if-you-don-t-run-your-own-llm-proxy-how-do-you-know-whats-going-on--3sryq76zv","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/if-you-don-t-run-your-own-llm-proxy-how-do-you-know-whats-going-on--3sryq76zv"},"datePublished":"2026-10-19T02:56:56.192Z","dateModified":"2026-10-19T08:59:10.769Z","description":"A developer argues that relying on AI providers' built-in usage dashboards is insufficient for real engineering decisions, since those panels lack time-series...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b8a307c45bf75ad07c3c4a73a182089b?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/b8a307c45bf75ad07c3c4a73a182089b?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hacker News","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Hacker News","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/hn","url":"https://daily.dev/sources/hn"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/if-you-don-t-run-your-own-llm-proxy-how-do-you-know-whats-going-on--3sryq76zv","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,observability,self-hosting","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hacker News","item":"https://daily.dev/sources/hn"},{"@type":"ListItem","position":3,"name":"If you don't run your own LLM proxy, how do you know whats going on?"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/if-you-don-t-run-your-own-llm-proxy-how-do-you-know-whats-going-on--3sryq76zv#faq","mainEntity":[{"@type":"Question","name":"Why would I want to run my own LLM proxy instead of using the provider's built-in analytics dashboard?","acceptedAnswer":{"@type":"Answer","text":"Provider analytics panels typically show surface-level usage stats, not engineer-grade time-series token data. A self-hosted proxy gives you a single endpoint across multiple providers and models, logs of which API key called what model and when, virtual API keys so you don't scatter one main key across agents and tools, full infrastructure control, and the ability to switch providers quickly without redoing setup. Developers weighing proxy setups for LLM observability can compare approaches and tooling discussions on daily.dev."}},{"@type":"Question","name":"What are the benefits of using virtual API keys with an LLM proxy?","acceptedAnswer":{"@type":"Answer","text":"Virtual API keys let you avoid littering a single main provider key across your work, agents, and harnesses. By issuing separate virtual keys per use case through a self-hosted proxy, you gain per-key logs of which client called which model and when, along with better security isolation and easier auditing than sharing one key everywhere. Teams managing multiple AI agents and tools can track key-management best practices via daily.dev."}}]} ```