Ask HN: How much money do you spend monthly on subscriptions for AI models?
Monthly total for all AI model subscriptions for personal use, with a breakdown of which models you use and for what purpose (coding side projects, AI assistants, etc.).
You can also mention how much you spend monthly on AI tools for professional work that your company pays for.
I spend $300 a month for Claude Max $200 and Google AI Ultra $100. Google Meets with live translation feature is a must-have for me, because I communicate with international clients almost everyday. It also gives me $40 monthly Google cloud credit where I can use it to test my client projects or my open source projects. Claude is my main tool for design, coding, and almost everything.
For me, the biggest pain point was actually having multiple AI subscriptions just to use different models for different things. I’m working on Talkory.ai, so I naturally use several models—GPT, Claude, Gemini, Grok, Perplexity, etc.—for things like coding, research, content and brainstorming. Individually, each subscription makes sense, but collectively the monthly cost adds up pretty quickly.
That’s one of the reasons I started building Talkory.ai: instead of paying for and switching between multiple AI tools, you can access and compare multiple models in one workspace and see which one gives the best answer for your particular task.
So rather than thinking “Which AI subscription should I pay for?”, I’m working on “Can I get the best model for each task without maintaining 5–6 separate subscriptions?”
I spend $300 per month on Claude's Pro subscription (20x tier), split with my cousin as we work as duo. It's pretty good and we never had to switch to another AI since we mostly code and Claude is pretty much the cream of the crop right now.
$0. but I use at least $8,000 worth of monthly tokens.
I used to have:
$200 Claude-max
$10 github copilot
$20 chatgpt
$20 Cursor
---
then, I spent roughly $5000 in tokens for a project which turned out to be a dud.
Figured, inference wasn't going anywhere as its now 80% of my day. So, I built a 4x DGX Spark cluster and put GLM5.2 on it. Then put a few qwen models for vision and embedding purposes on my original DGX Spark and completely went local. Tron, my Hermes agent, uses agent storming to orchestrate openCode agents and hermes subagents to handle most of my daily tasks.
For fun, I reworked a public Github vLLM Dashboard to track my usage and cloud cost avoided.
original dashboard was by github.com/niklasfrick but I extended it.
github.com/Wpnx330/spark-dashboard
---
For my day job I spend about $1000 + $20 a month in cursor costs.
I spend $20 for Claude, $10 for Github Copilot (thinking about cancelling because I'm using pi with openrouter api a lot lately), ~$30 for openrouter api.
Currently, I'm looking at Opencode Go for Chinese models. Can anyone compare Opencode Go and Openrouter if my use case is coding with K3 and GLM?
I have subscriptions for ChatGPT Plus and Claude Pro, so about $40/month. I use them for coding mostly, switching between them frequently and heavily. On the occasional times I run out of tokens I'll go outside and take a walk or spend time with family.
Been doing it this way for ~the last 6 months or so.
I do also use the API's for GPT and Gemini for some side projects, but the spend there is highly variable. It rarely exceeds $100 in a given month.
I used to spend 80$ per month on Claude pro while building my website + extension. However, once i launched the website and my extension i cancelled. I just feel i get what i need in the free plans of chatgpt, Gemini, and Claude.
Just $20 a month to Anthropic. I've run into having to step away from the AI projects for a bit here and there due to burning the 5-hour window in 3 hours a handful of times. But never ran out of the weekly usage with days to go.
Previously $0 using local models, then $10/mo with GitHub Copilot, cancelled that when it became a bad deal, now $10/mo with OpenCode Go.
I haven't seen a better deal than OpenCode Go yet. You get more in token spend than with using providers directly due to some volume discounts they pool.
exact same path as me. I'm still stuck with 5 more months of Copilot annual subscription, but it was nerfed so badly that I just wrote it off completely. OpenCode go in VS Code is fantastic.
Yep. Exact same take here. I have a private harness I actually integrated GitHub Copilot Chat support with, but you lose an "entire months" credits with any nominal task.
They knew what they were doing and wanted to kick people off.
Thank you for sharing your experience, too. I suspect a lot of us all took this path.
AMD 395+ w/128Gb personal use; work bought two older 48GB cards, and another framework 395+; all before the memory cartel. The my work machine was upgraded with blackwell 72gb.
All in, thats about 40k.
The amds act as regular pcs so not single purpose. The 2x 48gb will drive some RAG and backup IT admin. The 72gb will develop software.
The 395+ can run the 3.5 qwen A10B model at reasonable rates so using deer flow it can create arbitrary research reports and create green field projects like a cross browser extension scaffold.
Everytging im working on is in a niche that can benefit from custom software but cant invest in it. Its mostly about self resiliency rather than privacy.
If something happens to me, the AI can train anyone on the uses, so we are solving BUS factors rather than anything else.
Nice! I haven't been able to convince work to buy inference hardware. would love to get my hands on a 72gb card. Qwen3.8-27b just dropped and it would scream on a PRO6000.
run a 2-person custom dev agency and we just shipped our own saas, so ai is basically our co-founder at this point.
i currently maintain active subscriptions for claude pro and google ai pro. honestly, the roi is completely undeniable when you factor in how much time it saves us on boilerplate code, debugging, and drafting initial architectures. we also sprinkle in some midjourney and elevenlabs for asset generation when we need it.
for a small team trying to ship fast, the $40-$60/mo we spend on core llms is by far the cheapest overhead we have.
That’s one of the reasons I started building Talkory.ai: instead of paying for and switching between multiple AI tools, you can access and compare multiple models in one workspace and see which one gives the best answer for your particular task.
So rather than thinking “Which AI subscription should I pay for?”, I’m working on “Can I get the best model for each task without maintaining 5–6 separate subscriptions?”
I used to have: $200 Claude-max $10 github copilot $20 chatgpt $20 Cursor
---
then, I spent roughly $5000 in tokens for a project which turned out to be a dud. Figured, inference wasn't going anywhere as its now 80% of my day. So, I built a 4x DGX Spark cluster and put GLM5.2 on it. Then put a few qwen models for vision and embedding purposes on my original DGX Spark and completely went local. Tron, my Hermes agent, uses agent storming to orchestrate openCode agents and hermes subagents to handle most of my daily tasks.
For fun, I reworked a public Github vLLM Dashboard to track my usage and cloud cost avoided. original dashboard was by github.com/niklasfrick but I extended it. github.com/Wpnx330/spark-dashboard
---
For my day job I spend about $1000 + $20 a month in cursor costs.
Currently, I'm looking at Opencode Go for Chinese models. Can anyone compare Opencode Go and Openrouter if my use case is coding with K3 and GLM?
Been doing it this way for ~the last 6 months or so.
I do also use the API's for GPT and Gemini for some side projects, but the spend there is highly variable. It rarely exceeds $100 in a given month.
Mostly for side projects. I use GPT/Opus for planning, while Auto handles the implementation and the majority of the work.
At work, I have no life. I don’t have visibility into the cost.
I haven't seen a better deal than OpenCode Go yet. You get more in token spend than with using providers directly due to some volume discounts they pool.
They knew what they were doing and wanted to kick people off.
Thank you for sharing your experience, too. I suspect a lot of us all took this path.
Spent tons on hardware. Time is money and FOMO is a waste of money.
All in, thats about 40k.
The amds act as regular pcs so not single purpose. The 2x 48gb will drive some RAG and backup IT admin. The 72gb will develop software.
The 395+ can run the 3.5 qwen A10B model at reasonable rates so using deer flow it can create arbitrary research reports and create green field projects like a cross browser extension scaffold.
Everytging im working on is in a niche that can benefit from custom software but cant invest in it. Its mostly about self resiliency rather than privacy.
If something happens to me, the AI can train anyone on the uses, so we are solving BUS factors rather than anything else.
Personal use is much greater than use for company purposes.
$200 claude max through work -- pricing may be different, they don't show me the invoices
Personal: $19.99 on Google AI Pro
Zero at work.
i currently maintain active subscriptions for claude pro and google ai pro. honestly, the roi is completely undeniable when you factor in how much time it saves us on boilerplate code, debugging, and drafting initial architectures. we also sprinkle in some midjourney and elevenlabs for asset generation when we need it.
for a small team trying to ship fast, the $40-$60/mo we spend on core llms is by far the cheapest overhead we have.