Cycling through all six. Tap any point to stop.
Measured on six things
Compare GPUs, estimate local AI memory needs, and explore used asking prices.
Usability
63Search, tabs, and filters are easy to find, and docs/pricing FAQs are thorough. But the table throws TOPS and FP16 TFLOPS at newcomers with little in-page explanation, and the only strong CTA is a Log in button.
Screenshots
1 / 1
About this app
VRAMGlass is a free website for GPU comparison and local AI hardware planning. Compare specifications, explore formula-based memory estimates for AI models, and see used GPU asking-price references for supported markets. Asking prices come from listings and are not completed sales. An optional remote MCP service is available for supported clients.
Key features
- 1Live GPU price tracking across eBay, Xianyu, and Amazon Japan with 24h change and 30-day sparklines.
- 2Local LLM GPU ranking table with VRAM, bandwidth, FP16 TFLOPS, INT8 TOPS, power, and ecosystem tags.
- 3Formula-based VRAM estimator for open-weight models like Qwen, DeepSeek, and Kimi.
- 4Remote MCP server and REST API for pulling GPU prices and VRAM data into AI tools.
- 5Free tier with 50 calls/day; Pro at $7.17/month with 2,000 calls/day and 100 price alerts.
Target customer profile
Drawn from the product itself, not from a survey.
Scout Review
Written by AI from measured evidence, scored out of 100.
VRAMGlass is a fast, data-first GPU comparison and local-LLM hardware planning tool with a genuinely well-built ranking table, live multi-market price data, a VRAM estimator, and an MCP server for AI tooling. The free tier is real, the $7/month Pro tier is cheap, and the docs are unusually thorough for an indie product. Weaknesses are execution details: tiny table type, an ad banner above the primary data, inconsistent vendor logos, and an expert-oriented interface that assumes you already know what TOPS means. Security is solid but unaudited; accessibility is strong with two fixable failures. Growth hinges on whether the MCP integration gets adopted by AI tooling — a smart wedge in a category with real headroom.
Measure by measure
Usability
63Search, tabs, and filters are easy to find, and docs/pricing FAQs are thorough. But the table throws TOPS and FP16 TFLOPS at newcomers with little in-page explanation, and the only strong CTA is a Log in button.
Design
71Data-first layout with a disciplined ranking table, aligned numerics, sparklines, and restrained typography. Docked for tiny table type, an ad banner above the primary data, and inconsistent vendor logos.
Speed
88Lighthouse mobile 99/100 (LCP 2.0s, CLS 0.001, TBT 17ms) and desktop 100/100. The site is genuinely fast and stable under load.
Security
64HTTPS enforced with HSTS, plus X-Content-Type-Options, Referrer-Policy, SPF, and DMARC. Missing CSP, frame protection, and Permissions-Policy; no published audits or bug bounty found.
Accessibility
79Lighthouse accessibility 96/100 with multi-language UI and currency switching. Two failing audits: insufficient touch target sizing and visible labels without matching accessible names.
Growth
68Real, specific problem (local-LLM hardware buying) with concrete wedges: live multi-market price data, VRAM estimator, and an MCP server for AI tools. Category has headroom as local inference grows.
Improvement areas
- Table typography is too small and rows too tight for comfortable scanning at normal viewing distance.
- Ad banner sits above primary data and competes with the ranking table.
- No CSP, frame protection, or Permissions-Policy; no published audits or bug bounty.
- Two Lighthouse a11y failures: touch target sizing and label/accessible-name mismatches.
Design & Usability
VRAMGlass looks like a tool built by someone who actually uses it. The ranking table is the star: consistent rows, right-aligned numerics, inline sparklines, and green/red deltas that let a buyer scan 40+ GPUs quickly. Typography is restrained and the data carries the visual weight. The problems are execution details: table type is small enough to squint at, the Vast.ai ad banner sits directly above the primary content, and vendor logos are inconsistently sized. Usability is split — search, tabs, and filters are easy to find, and the docs and pricing FAQs are unusually thorough for an indie tool. But the table assumes fluency in TOPS and FP16 TFLOPS, and the only strong CTA is a Log in button. Newcomers get a wall of numbers; experts get exactly what they want.
Speed & Security
Speed is the standout: Lighthouse mobile 99/100 with LCP 2.0s, CLS 0.001, and TBT 17ms, plus a perfect 100 on desktop. Whatever stack this is, it's doing the right things. Security is competent but not exceptional. HTTPS is enforced with HSTS, and X-Content-Type-Options, Referrer-Policy, SPF, and DMARC are all present — 8 of 12 checklist points. Missing are CSP, frame protection, and Permissions-Policy, and there's no published SOC 2, ISO, pen-test, or bug bounty. For a free tool with an optional $7/month API tier, that's a reasonable posture, but anyone wiring the MCP server into their own AI tooling should know there's no third-party audit behind it.
Accessibility & Growth
Accessibility lands at 96/100 on Lighthouse with two real failures: touch targets too small and visible labels without matching accessible names. The multi-language UI (English, Simplified and Traditional Chinese, Japanese) and currency switching are genuine inclusivity wins. Growth is the interesting question. The problem is real and specific — people building local LLM rigs need to know what a used GPU costs and how much VRAM a model needs — and the wedges are concrete: live eBay/Xianyu/Amazon.jp price data, a VRAM estimator, and an MCP server that plugs into Claude Code, Cursor, and similar tools. That MCP angle is the actual bet: if AI assistants start calling VRAMGlass for GPU prices, it becomes infrastructure rather than a destination site. The category has headroom as local inference keeps growing, but the domain is only ~0.2 years old, so this is thesis, not traction.
Conclusion
If you're speccing a local LLM rig and tired of tab-hopping between eBay, spec sheets, and VRAM calculators, VRAMGlass collapses that workflow into one page. The free tier is genuinely useful, the Pro tier is priced like a coffee, and the MCP server is the kind of integration that could quietly become load-bearing for AI-assisted hardware shopping. It's early — the domain is months old — so treat the price data as a reference, not gospel, and don't expect hand-holding on what the columns mean. But as a fast, honest, data-first utility, it earns a bookmark.
Suggestions for improvement
- Add a short glossary or tooltip layer for TOPS, FP16 TFLOPS, and bandwidth so newcomers aren't bounced by the table.
- Move the Vast.ai ad banner below the stat cards or into a sidebar so it stops competing with the primary ranking data.
- Publish a security page or at least a CSP and Permissions-Policy to close the remaining header gaps.
- Fix the two Lighthouse a11y failures: enlarge touch targets and align visible labels with accessible names.
Alternatives comparison
Named competitors, point by point. Nobody paid to appear here or to be left out.
| Live used-GPU asking prices across multiple marketplaces | Yes — eBay, Xianyu, Amazon.jp with 24h change and 30-day sparklines | Not specified in provided details |
|---|---|---|
| VRAM estimator for open-weight LLMs | Yes — formula-based estimates for models like Qwen, DeepSeek, Kimi | Not specified in provided details |
| MCP server / API for AI tool integration | Yes — remote MCP server plus REST API, free tier 50 calls/day | Not specified in provided details |
| Pricing model | Free tier; Pro $7.17/month or $86/year, one-time packs, no auto-renew | Not specified in provided details |
| Multi-language and multi-currency support | English, Simplified/Traditional Chinese, Japanese; USD/CNY/JPY/TWD/EUR | Not specified in provided details |
Comparing options? See VRAMGlass alternatives, scored side by side
More Development & Engineering apps
View all Development & Engineering
Vibe Coding Agent Skills
92A free, open-source platform offering 110 AI agent skills...

CreateMyCoin
92A no-code Solana token creator that deploys SPL tokens in...

CC Switch
91Open-source desktop app for switching AI coding providers...

TaxID
90Fast EU VAT validation API for developers, covering 27 EU...

DemoVexa
90Turn your code into professional product demo videos with AI-generated scripts, animated UI, and natural voiceovers.

FontPlop
89Browser-based font viewer & tester for TTF, OTF, WOFF, WO...

deed.page
89Permanent web hosting for AI agents, no human step

Webfetch
89Web scraping API that turns any URL into HTML, text, Mark...

Denpex
89AI-powered GPU failure diagnosis for distributed ML training

BeartIMAGE
88BeartIMAGE is a free, web-based image processing platform engineered for fast, bulk photo editing and conversion directly in your browser.

Claw Messenger
88Claw Messenger is an iMessage API service that gives AI a...

DataHabibi
88AI-powered Dubai real estate data platform with 12M+ DLD ...
A platform dedicated to providing unbiased reviews of newly launched applications, analyzing everything from their features to their full potential.
info@scoutforge.net© 2026 Scoutforge. All rights reserved.