Developer docs
Bring VRAMGlass data into your AI tools or embed it on your website. Both are free; please credit the source when you cite it.
MCP setup
Use MCP to bring live GPU prices into AI tools like Grok Bot, Claude Code and Cursor. It's free and needs a VRAMGlass account. See the VRAMGlass MCP page for what you can ask.
https://mcp.vramglass.com/mcpSetup steps
- Sign up or log in to VRAMGlass
It's free. All you need is an email address.
Sign up - Create an API key
Open Your account → MCP connections, click Create key, give it a name (for example "Grok") and choose how long it stays valid. The key is shown only once, so copy it right away.
Open MCP connections - Send it to your AI
Copy the text below into your AI chat and replace YOUR_KEY with the key you just copied. The AI adds VRAMGlass by itself:
Setup promptPlease add a remote MCP server: Name: VRAMGlass URL: https://mcp.vramglass.com/mcp Header: Authorization: Bearer YOUR_KEY Once it's added, try it by checking what an RTX 4090 costs right now.
This works with Grok Bot, Claude Code, Codex, Manus, WorkBuddy and any other AI that can change its own settings.
- Ask it something
For example, "How much is an RTX 5090 right now?" or "How much VRAM do I need for Qwen3.8 27B?" The AI answers with VRAMGlass data and links to the source.
Manual configuration
If your tool asks you to fill in the settings yourself, it needs just two things: the server URL and the header Authorization: Bearer YOUR_KEY. Choose HTTP (Streamable HTTP) as the transport.
claude mcp add --transport http vramglass https://mcp.vramglass.com/mcp --header "Authorization: Bearer YOUR_KEY"
{
"mcpServers": {
"vramglass": {
"url": "https://mcp.vramglass.com/mcp",
"headers": { "Authorization": "Bearer YOUR_KEY" }
}
}
}{
"mcpServers": {
"vramglass": {
"url": "https://mcp.vramglass.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_KEY"
}
}
}
}Sign in with OAuth
Connectors in Claude (web and desktop) and ChatGPT only support OAuth and can't take a key. Any other MCP tool that supports OAuth can connect the same way, with no key needed:
- Add an MCP server in your tool and paste the server URL, without any header. In Claude, add a custom connector under Settings → Connectors; in ChatGPT, add it under Settings → Apps in developer mode.
- The tool opens the VRAMGlass login page. Log in with an email code or Google; if you don't have an account, one is created.
- The consent page shows the app's name, the address it will return to and what it can do. Make sure you started the connection yourself, then click Allow. App names are provided by the apps themselves and aren't verified by VRAMGlass.
- You're sent back to your tool, connected. Just start asking.
- An authorized app can only read public price and spec data. It can't see your email, price alerts or watchlist, and it can't change anything in your account.
- The tool renews access on its own, so you don't have to keep logging in. After 30 days without use, or 180 days after you authorized it, you'll need to authorize it again.
- Authorized apps are listed under Your account → MCP connections → Authorized apps. Click Disconnect and the app stops working within 30 seconds.
Keeping your key safe
- A key can only read public price and spec data. It can't see your email, price alerts or watchlist, and it can't change anything in your account.
- Use one key per tool so you can turn each one off separately. Each account can have up to 5 keys, and you choose how long each one lasts when you create it, up to one year.
- Your key stays in your chat history with the AI. If it leaks or you no longer need it, revoke it under MCP connections and it stops working within 30 seconds.
Available tools
Once connected, the AI picks the right tool by itself. You don't need to remember these names.
get_prices- Current prices for GPUs or memory on eBay, Xianyu and Amazon Japan: median, price range and listing count
get_price_history- Daily price history, from the last 7 days up to a year
get_price_index- The GPU price index and its price tiers
search_hardware- Find GPUs and memory by name
search_models- Find LLMs and their versions by name
models_for_gpu- Which models a GPU can run, and how long a context fits
gpus_for_model- How much VRAM a model needs, which cards can run it and roughly how fast
recommend_build- PC builds for a model (GPU, CPU, motherboard, RAM, PSU) with an optional budget, prices and buy links; GPU options instead when the model fits entirely in VRAM
get_ebay_deals- The eBay deals shown on GPU and memory pages: the same rule-filtered listings, up to six
get_rent_prices- Hourly rental prices on RunPod, Vast.ai and Compshare
Usage
It's free. If you call it too often you'll be rate-limited briefly, then it recovers on its own. For heavy use, email hello@vramglass.com.
Widget
Embed a GPU price trend card in your blog or forum with one snippet. The numbers update together with the site. Open source under the MIT license.
<script src="https://vramglass.com/widget/v1/price-card.js" async></script> <vg-price-card gpu="rtx-3090" market="ebay"></vg-price-card>
gpu- The GPU: the last part of its page URL, for example rtx-3090
market- Marketplace: ebay, xianyu (Xianyu) or amazon_jp (Amazon Japan)
currency- Display currency: USD, CNY, JPY, TWD or EUR. Leave it out to use the marketplace's own currency
locale- Language: en, zh, zh-tw or ja. English if left out
theme- light, dark or auto (follows the system; the default)
Data and citation
- Prices are the median asking price of listings currently for sale, not sold prices. eBay, Xianyu and Amazon Japan are counted separately.
- Prices update every 6 hours. A round with too few valid listings doesn't publish a price.
- When you cite the data, credit "VRAMGlass", link back to the relevant page, and state the marketplace and date of each price.
- Buy and rental links carry VRAMGlass referral parameters. The price you pay stays the same.