Stack Cost AI

Agentshobby stack.

For each requirement below, pick the option that fits your build — recommended first, then free and cheaper alternatives — or skip what your project doesn't need. Tap the info icon next to any requirement to see why it matters.

LLM Inference API / Foundation Model Access

Pricing & free-tier limitsOpenAI: Free $5 credit; $2.50-$5/1M input, $10-$15/1M output (4o). Anthropic: $3/1M in $15/1M out (Sonnet 3.5). Groq: Free 100 req/min 14.4k/day, then payg $0.59/1M. Together: $0.20/1M tokens. Overage: pure payg, no min.

Agent Orchestration Framework

Pricing & free-tier limitsLangGraph OSS free unlimited; Cloud Free 5k runs/mo, $39/mo Dev 10k runs $0.01/run overage, Pro $399/mo 100k runs. CrewAI OSS free; Cloud $29/mo 1k runs $0.03/run overage. AutoGen free.

LLM API Gateway (Multi-model proxy)

Pricing & free-tier limitsLiteLLM OSS free unlimited; Cloudflare AI Gateway Free 1M req/mo cached logging free. Portkey Free 10k req/mo, $100/mo 100k req, $0.001/1k overage. Helicone Free 10k logs/mo then $20/mo. Caching saves 30-50% LLM cost.

Memory / Vector Database for Agent Memory

Pricing & free-tier limitsPinecone Free 100k vectors 1 index; Starter $70/mo supports 1M vectors $0.33/GB storage + $0.33/1M reads. Qdrant Free 1GB RAM cluster (5M vectors @ 768d), $25/mo 2GB 10M vectors, overage $0.10/GB storage. Mem0 Free 1k memories, Cloud $49/mo 50k.

State Management / Checkpointing / Session Store

Pricing & free-tier limitsUpstash Redis Free 10k cmds/day 256MB max 1k conn; PayG $10/mo 100k cmds/day 1GB, $0.20/100k cmds overage. Redis Cloud Free 30MB, Essentials $5/mo 100MB $0.20/100k ops. Neon Free 0.5GB storage 3 projects, $20/mo Scale 10GB.

Sandbox Execution Environment (Code execution)

Pricing & free-tier limitsE2B Free 100 hours/mo CPU sandbox, 30 concurrent, 2GB RAM; Starter $25/mo 1000 hours + 10 vCPU, $0.03/vCPU/hr overage + $0.10/GB RAM hr. Modal Free $30 credit/mo (~30 E2B hrs), then $0.000032/vCPU/sec (~$0.12/vCPU/hr). Daytona OSS free 5 workspaces.

Tool/Function Calling Infra (MCP Registry)

Pricing & free-tier limitsComposio Free 500 tool calls/mo 3 MCP custom servers; Pro $49/mo 10k calls $0.005/call overage $0.10/active connection. MCP OSS unlimited. Toolhouse Free 1k calls/mo $25/mo 10k. Zapier 750 tasks free.

Database - Relational (Agent Metadata, Runs)

Pricing & free-tier limitsSupabase Free 500MB DB 50k MAU auth 2GB egress 1GB file; Pro $25/mo 8GB DB 50GB egress $0.0035/hr compute, storage overage $0.021/GB. Neon Free 0.5GB, 10k compute hours free then $0.16/CU hr. Turso Free 9GB, $29/mo 24GB.

Monitoring & Tracing (Agent Traces)

Pricing & free-tier limitsLangSmith Free 5k traces/mo 14d retention; Plus $39/mo 10k traces, $0.005/trace overage. Langfuse Free Cloud 50k observations/mo 30d, $39/mo 100k obs $0.003/observation overage. Helicone Free 10k req/mo. Self-host OSS free unlimited.

Secrets Management for Tool Credentials

Pricing & free-tier limitsInfisical OSS free unlimited self-host; Cloud Free 5 projects 100 secrets; Pro $12/seat/mo. Doppler Free 5 users 25 configs; Team $18/user/mo unlimited. AWS Secrets $0.40/secret/mo + $0.05/10k calls. Vault HCP free 25 secrets.

Logging & Observability Platform

Pricing & free-tier limitsGrafana Cloud Free: 50GB logs/mo 14d retention, 50GB traces, 10k metrics; Pro $8/user/mo + $0.50/GB logs ingestion $0.30/1k hours traces overage. Datadog Free 1-day logs; Pro $15/host/mo APM + $0.10/GB logs. SigNoz OSS free unlimited self-host; Cloud $49/mo 100GB.

Options and prices come straight from our research sheets for a hobby agents project. Prices are estimates and change often — always confirm on the provider's page before committing.

How this hobby agents checklist works.

Each requirement below is something a hobby agents build typically needs. Pick one of the four researched options — recommended, free, cheaper or paid — add your own with "Other", or skip the requirement if your project doesn't need it. Nothing is mandatory; the plan on the right tracks what you've decided so nothing gets forgotten.

Your picks are saved in this browser automatically, so you can come back anytime. Options are researched per build level and refreshed as vendors change their plans — always verify details on the provider's page before committing.