We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/markster-exec/mcp-unicorn'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
---
id: cloudflare-hackathon-angle
title: "Cloudflare Hackathon Angle — Workers AI, AI Gateway, MCP"
type: research
status: active
owner: ivan
created: 2026-02-19
updated: 2026-02-19
tags: [hackathon, research, sponsors, cloudflare]
---
# Cloudflare Hackathon Angle
## Workers AI
Serverless AI inference on Cloudflare's global GPU network (200+ cities). 50+ models: text gen (Llama 4, GPT-OSS, Qwen3, Gemma 3, Mistral), image (FLUX.2), speech (Whisper, Deepgram), embeddings (BGE, Qwen3-embedding).
### Pricing
- **Neurons** (GPU compute units): $0.011 per 1,000 Neurons.
- **Free**: 10,000 Neurons/day (resets 00:00 UTC).
- Example: Llama-3.1-8b ~$0.15-0.83/M tokens. FLUX image: pennies per tile. Whisper: ~$0.0005/audio minute.
- No idle costs; pay-per-inference.
## AI Gateway
Reverse-proxy/observability for ANY AI provider (Workers AI, OpenAI, Anthropic, etc.).
- **Observability**: Dashboard analytics (requests, tokens, costs, latency, errors). Full request/response logging.
- **Caching**: Edge caching (up to 90% latency/cost reduction on repeat prompts). Custom rules, TTL.
- **Rate Limiting**: Per-user/IP/key, sliding/fixed windows. Prevents abuse.
- **Retry & Fallback**: Auto-retry on errors; define model/provider fallbacks.
Free tier: ~100k total logs/month. Paid: 1M+.
## D1 + Vectorize
- **D1**: Serverless SQLite. Time Travel (point-in-time restore). Free: 10 DBs, 500 MB/DB.
- **Vectorize**: Globally distributed vector DB. Median query latency ~31ms. RAG: generate embeddings → store → query → feed to LLM.
## MCP Support
Cloudflare has deep MCP support:
- Build/deploy MCP servers on Workers (HTTP endpoints exposing tools/prompts/resources).
- **Agents SDK**: Goal-driven agents that reason, call tools, schedule tasks on Workers (stateful via Durable Objects).
- Managed/remote MCP servers from Cloudflare: AI Gateway, AI Search, Docs.
- Full catalog in GitHub (mcp-server-cloudflare).
## Workers Free Tier Limits
- Requests: 100k/day.
- CPU Time: 10ms per invocation (Free); up to 5 min on Paid.
- Memory: 128 MB.
- Paid ($5/mo min): 10M requests/month, much higher limits.
## What Would Impress Cloudflare
Full end-to-end Cloudflare stack: Pages + Workers + Workers AI + Vectorize + D1 + AI Gateway + Agents SDK + MCP server + Durable Objects + R2.
**Dream project**: AI voice/customer-service agent with MCP tools + RAG over docs; or personalized recommender with Vectorize + image gen; or secure agent that edits configs via MCP.
**Previous sponsor prizes**: Up to $250K in credits (TreeHacks 2026!), swag, hardware.
**Pro tips**:
- Use MCP + Agents (new/hot).
- Demo AI Gateway dashboard live (caching savings, logs).
- Combine Vectorize + D1 + Workers AI for RAG.
- Ship fast, document the stack, show metrics/latency.
Sources: developers.cloudflare.com, Cloudflare blog, DevWeek announcements, hackathon sponsor pages.