Pricing built for the network

Free

For side projects and experimentation. Includes FunctionFly AI.

• ✓25K requests / month
• ✓3 published functions
• ✓FunctionFly AI (Mercury 2): 500 req/week · 166/day
• ✓BYOK AI (any provider, rate limits apply)
• ✓L1 verification
• ✓Community support
• ✓7-day execution logs
• ✓State Fabric: 1 object · 10 MB · 100K ops/mo
• ✓Vault: 25 secrets
• ✓500 agent memories
• ✓Studio (1 project/user)

Starter

For side projects and MVPs.

• ✓500K requests / month
• ✓10 published functions
• ✓FunctionFly AI: 420 req/week · 140/day
• ✓BYOK AI: 100K calls/mo · 100 calls/min
• ✓10 AI agents (BYOK)
• ✓L1–L2 verification
• ✓Email support
• ✓30-day execution logs
• ✓Zero-knowledge Vault (500 secrets)
• ✓State Fabric: 50 objects · 5 GB · 10M ops/mo
• ✓5,000 agent memories
• ✓Function DNA
• ✓Always-on: 50h/month + 256MB context pool
• ✓Sandbox daily cap: 1 h wall time
• ✓Studio (5 projects/user)

Professional

For growing teams and production SaaS.

• ✓2.5M requests / month
• ✓25 published functions
• ✓FunctionFly AI: 2,100 req/week · 700/day
• ✓BYOK AI: 1M calls/mo · 500 calls/min
• ✓100 AI agents (BYOK)
• ✓L1–L3 verification
• ✓Email support
• ✓90-day execution logs
• ✓Zero-knowledge Vault (5,000 secrets)
• ✓30-day Time Machine
• ✓State Fabric: 500 objects · 50 GB · 100M ops/mo + Hot Cache
• ✓25,000 agent memories
• ✓Function DNA
• ✓Consciousness (basic)
• ✓Studio (25 projects/user)
• ✓Always-on: 200h/month + 1GB context pool
• ✓Sandbox daily cap: 5 h wall time

Enterprise

For large-scale apps, compliance, and custom trust policies.

• ✓25M requests / month
• ✓Unlimited functions
• ✓FunctionFly AI: 8,400 req/week · 2,800/day
• ✓BYOK AI: 5M calls/mo · 2,000 calls/min
• ✓500 AI agents (BYOK)
• ✓L1–L4 verification
• ✓Priority + SLA
• ✓1-year execution logs
• ✓Zero-knowledge Vault (1M secrets)
• ✓90-day Time Machine + reconciliation
• ✓RBAC + Secret sharing
• ✓Unlimited State Fabric (500 GB · 1B ops/mo) + all add-ons
• ✓100,000 agent memories
• ✓Function DNA
• ✓Advanced Consciousness
• ✓Studio (unlimited projects)
• ✓Always-on: 1,000h/month + 10GB context pool
• ✓Sandbox daily cap: 24 h wall time
• ✓MicroVM Enterprise: 100 concurrent · 2 vCPU · 512MB–2GB

MicroVM Enterprise

Heavy MicroVM workloads (500 concurrent, 16 vCPU, 8 GB).

• ✓Everything in Enterprise
• ✓500 concurrent MicroVMs (5× standard)
• ✓16 vCPU / 8 GB per MicroVM (4× standard)
• ✓$50/mo included compute budget
• ✓Discounted overage rates
• ✓60 s default / 10 min max timeout

Enterprise SLA

Five-nines SLA + dedicated SLA engineer + 15-minute response.

• ✓Everything in Enterprise
• ✓99.999% uptime SLA (five nines)
• ✓15-minute response time guarantee
• ✓Automatic SLA credits for violations
• ✓Priority incident handling
• ✓Dedicated SLA engineer

Agent Enterprise

Unlimited AI agents + 42K FunctionFly AI req/week.

• ✓Everything in Enterprise SLA
• ✓Unlimited AI agents & concurrency
• ✓FunctionFly AI: 42,000 req/week · 14,000/day
• ✓250,000 agent memories
• ✓MicroVM Enterprise: 50 concurrent · 16 vCPU
• ✓Unlimited Time Machine + incident insurance

Do prices include tax?

Listed prices are in USD before applicable taxes. Your invoice will reflect tax rules for your billing address.

What's the difference between BYOK and FunctionFly managed AI?

BYOK (Bring Your Own Key): you wire your OpenAI, Anthropic, Gemini, Groq, or any OpenAI-compatible endpoint. FunctionFly never sees your key on the wire (Vault ZK mode) and the call is billed directly by your provider. FunctionFly managed AI: FunctionFly routes through Cloudflare Workers AI and Inception Mercury 2, paying the upstream cost out of platform margin; you draw from the per-tier weekly quota (Free 500/wk, Starter 420/wk, Pro 2,100/wk, Enterprise 8,400/wk, Agent Enterprise 42,000/wk) or buy credits once you exceed it. Cache hits count as 1/10 of a quota unit, so the effective quota is ~10× on repeated prompts.

Why does Free include FunctionFly AI but paid tiers have lower weekly quotas?

Free uses Inception Mercury 2 (~30× cheaper per-call than Cloudflare Workers AI) for the diffusion-LLM rung. The 4.8× quota bump on 2026-09-02 was safe because worst-case COGS for the entire free tier stays under $1.20/week per user. Paid tiers default to Cloudflare Workers AI for higher reliability + more model options — Starter doesn't get the bump because it's still on Workers AI. Pro and above get the same Mercury 2 model-by-default once AIMODEL_ROUTER_PRIMARY_FUNCTIONS includes your function.

How does the Trust API and verification billing work?

Trust scores, attestations, and verification levels are built into every function — included in your plan. Usage-based components (extra verification drills, deeper audit trail) are metered separately so you only pay for what your agents and policies actually query.

What are always-on agents?

Always-on agents stay resident between requests, maintain persistent memory across sessions, and respond instantly without cold-start latency. Each plan includes a monthly budget of always-on hours and a shared context memory pool; usage beyond that is metered per-minute.

How does always-on billing work with BYOK?

Your LLM costs go directly to your provider via your own API key. The platform charges only for the always-on infrastructure—compute, memory, and orchestration—at $0.05–$0.10 per active minute (depending on plan), with idle time at 25% of that rate.

Can we upgrade or downgrade?

Yes. Move between platform tiers as your traffic and team change; agent plans can be adjusted when your concurrency and call volume need it. Mid-month upgrades include prorated credit for unused hours on the old plan. Annual plans save 20% — or 30% through Nov 30 with the FALL26 promo.

What's the difference between Enterprise, MicroVM Enterprise, Enterprise SLA, and Agent Enterprise?

Enterprise ($299/mo) is the base paid tier with 25M requests, 100 concurrent MicroVMs, and 99.99% SLA. MicroVM Enterprise ($399/mo) is for heavy container workloads — 500 concurrent MicroVMs (5× standard), 16 vCPU and 8 GB per MicroVM (4×), $50/mo included compute budget, and discounted overage rates. Enterprise SLA ($499/mo) upgrades the base Enterprise tier to a five-nines (99.999%) uptime SLA with a 15-minute response time guarantee and a dedicated SLA engineer. Agent Enterprise ($499/mo) is unlimited agents + 42K FunctionFly AI requests/week + 250K memories + all of the Enterprise SLA benefits.

What does Enterprise include?

Volume pricing, custom SLAs, security questionnaires, SSO, and tailored trust policies. Enterprise tier includes L1–L4 verification, 1-year execution logs, 90-day Time Machine with reconciliation, RBAC + Secret sharing, 100K agent memories, Advanced Consciousness (basic ML-driven insights + auto-fixes), 100 concurrent MicroVMs, 1,000 always-on hours/month, and a 10GB context pool. Add MicroVM Enterprise or Enterprise SLA upgrades as needed.

Is there a free trial for paid plans?

Starter and Professional come with a 14-day free trial — no credit card required to start. Enterprise, MicroVM Enterprise, Enterprise SLA, and Agent Enterprise are sold via our sales team (Contact sales) so we can scope SLAs, scale caps, and onboarding to your workload before the contract starts.

How does sandbox per-second billing work?

Every sandbox execution (python-light, node-lite, node22-typescript, typescript-wasm, ruby, FlyBox, CCP GPU pods) is metered per second of wall time. One row per execution is written to the ledger with kind='sandbox_seconds'; per-second resolution lives in the Prometheus histogram and the dashboard's Sandbox Usage panel. Daily per-tenant caps (Free 10 min, Starter 1 h, Pro 5 h, Enterprise 24 h) refuse new admissions when exceeded — in-flight sandboxes always finish. GPU pods (ccp.l4, ccp.a5000, ccp.a100, ccp.h100) are billed at a higher per-second rate to cover upstream provider costs.

What happens when my sandbox daily cap is hit?

New sandbox starts return a typed 'sandbox_daily_cap_exceeded' error. Currently-running boxes continue to completion — the cap is enforced at admission only. The optional SANDBOX_LOW_BALANCE_KILL_THRESHOLD env var enables hard-kill for ops-driven drain scenarios.

Do you offer a discount for non-profits, open-source projects, or startups?

Yes. FunctionFly for Startups offers up to $100K in credits (similar to Render's startup program). Open-source maintainers get a free Enterprise tier for qualifying projects. Non-profits get 50% off any paid tier. Contact sales with your details.

How do I redeem the Fall 2026 promo?

Click any signup link with ?promo=fall26 — the FALL26 code is auto-applied at checkout for a 30% annual discount (2 months free) through November 30, 2026. To stack an additional offer (DOUBLEUP, FLYAI2X, SANDBOX5H, STARTUP100, or REFERRAL), enter the bonus code in the promo field on the second checkout step. OSS-FREE is standalone and approved via the GitHub repo submission. Promos are not retroactive.