Pricing

Start free.
Scale as you grow.

Two ways to pay. Managed bills 1 Neuron per AI operation — flat and predictable. BYOK lets you bring your own provider keys and pay per-token, 10% below OpenRouter.

Free
$0 forever
500 Neurons / mo

For developers and validation

Start free
  • Chat completions
  • RAG (50 documents)
  • 1 agent
  • Open-source edge models
  • Community support
Most popular
Builder
$29 /month
10,000 Neurons / mo

For developers shipping AI features

Save $500+/mo vs. equivalent stack
Start building
  • All composable patterns
  • Unlimited RAG documents
  • Agents & workflows
  • Batch inference — 40% off
  • OpenAI + Anthropic models
  • Email support
Scale
$99 /month
75,000 Neurons / mo

For teams with production workloads

Scale up
  • Everything in Builder
  • BYOK provider keys
  • SSO + RBAC
  • MCP server
  • Composite profiles
  • 10 team seats
  • Priority support
Enterprise
Custom
Unlimited Neurons

For regulated industries and large deployments

Talk to us
  • Everything in Scale
  • SLA guarantee
  • Compliance docs
  • Custom domain
  • Deploy to your CF account
  • Dedicated engineering
Two ways to pay

Managed, or bring your own keys.

Default

Managed — pay per Neuron

1 Neuron = 1 AI operation, regardless of model, provider, or prompt size. We hold the provider keys and absorb token costs. Flat, predictable billing — see the plans above.

  • No token-counting, no per-seat math
  • Open-source edge models always free
  • 500 Neurons/mo free, then per-operation overage
Scale plan & up

BYOK — pay per token

Bring your own OpenAI or Anthropic keys (encrypted per tenant). Route through our gateway and pay your provider directly — always 10% below OpenRouter list, batch up to 40% off. Neureus charges only for operations.

BYOK token rates

10% below OpenRouter.
Batch up to 40% off.

When you bring your own keys, our gateway preprocesses every prompt — trim, dedupe, compress — so your provider bill is lower than routing raw through OpenRouter. Managed mode bills per Neuron instead (above).

Model OpenRouter BYOK via Neureus BYOK Batch You save
GPT-4o $5.00 / 1M tokens $4.50 / 1M $3.00 / 1M 10–40%
Claude Sonnet 4.6 $3.00 / 1M tokens $2.70 / 1M $1.80 / 1M 10–40%
Llama 3.3 70B $0.59 / 1M tokens Free Free 100%
Mistral 7B $0.07 / 1M tokens Free Free 100%

* Open-source models (Llama, Mistral, Qwen, DeepSeek) run free at the edge in both modes. Per-token rates apply to OpenAI and Anthropic models in BYOK mode only.

Cost calculator

Estimate your cost.

AI operations per month 10,000
Managed — your monthly cost
$29.00
Builder · all included
Builder — $29/mo

1 Neuron per operation. 500 free, then $3 per 1,000 Neurons (batch $1.80). We cover the provider token costs.

FAQ
How does BYOK beat OpenRouter on token cost?
In BYOK mode, every request passes through a 4-pass preprocessor before reaching your provider — Normalize → Structure → Trim → Compress (opt-in). Your provider bills for what the model actually processes, not the raw payload, so your per-token spend lands 10–30% below routing raw through OpenRouter. Batch mode (provider Batch APIs) adds another 40%. In Managed mode you don't see tokens at all — you pay 1 Neuron per operation and we absorb the provider cost.
What exactly is a Neuron?
1 Neuron = 1 AI operation. A chat turn, RAG query, agent step, workflow execution, or composite pattern call — each costs exactly 1 Neuron regardless of model, provider, or prompt size. No token-counting. No per-seat math. What you see on the calculator is what you pay.
Can I bring my own OpenAI or Anthropic keys?
Yes — on the Scale plan and above. BYOK stores keys encrypted per tenant. You pay your provider directly; Neureus charges only Neurons above your plan allocation — no markup on model cost.
What happens when I exceed my Neuron allocation?
Overage at $3.00 / 1,000 Neurons. Batch mode overage at $1.80 / 1,000 Neurons. Email alert at 80% of monthly allocation. Set a spending cap in Settings → Billing. Unused Neurons don't roll over.