▲ hostfleet /v2
The AI Infrastructure Review

Ship AI apps.
Everywhere.

Independent benchmarks of GPU inference hosts, edge runtimes, and the platforms where Lovable, Bolt, and v0 outputs actually live in production. No fluff, no "best of" filler — just real data from real deploys.

AI hosting → Edge runtimes →

// flagship

ai-hosting
Cold start latency showdown: 8 serverless GPU providers, 4 model sizes
Real cold-start numbers from Runpod, Modal, Fal, Baseten, Replicate, Beam, Banana, and SageMaker across 7B, 13B, 34B, and 70B models. Plus the weight-baking trick that changes the math.
deploy-ai-apps
We deployed the same Lovable app to 5 hosts. Here is what broke.
A real Lovable-generated SaaS deployed to Vercel, Netlify, Cloudflare Pages, Fly, and Render. Cold starts, build times, edge behavior, and what each host actually costs when an AI builder hands you the zip.
ai-hosting
$5 vs $500 GPU: Runpod, Modal, and Fal running Llama 3.3 70B
Three serverless GPU hosts, one 70B model, real throughput and cost per million tokens. Who wins depends on whether your traffic is bursty or steady.

// recent