Pricing
UClaw uses simple pay-as-you-go billing. You add funds to your account and spend them only on the tokens your agents consume. No subscription fees, no seat licenses, no hidden costs.
Billing
| Markup | 15% on raw LLM token costs (agentic runs only) |
| Free tier | $1.00 on signup |
| Minimum top-up | $10.00 |
How Billing Works
Your account balance is deducted per agent run based on the tokens used:
- Input tokens: The prompt, conversation history, and system instructions sent to the model
- Output tokens: The tokens generated by the model in response
- Cache reads/writes: Some models support prompt caching, which reduces costs on repeated context
Formula (agentic runs): amount charged = ((input_tokens × input_rate) + (output_tokens × output_rate)) / 1,000,000 × 1.15
The 1.15 multiplier is UClaw's 15% infrastructure fee and applies to agentic runs — durable agent sessions that include persistent state, real-time streaming, and tool execution. This fee covers the compute, memory, and orchestration infrastructure that keeps your agents running.
NOTE
The 15% markup applies to agentic runs only. If you use UClaw purely as a passthrough for plain text generation (no stateful agent, no tools), no infrastructure fee is added — you pay raw model token costs directly.
NOTE
We are transparent about our markup. Raw model costs come directly from the upstream AI providers. You can verify our prices against their public pricing pages — we add exactly 15% on agentic runs.
Free Tier
Every new account receives $1.00 automatically on signup. No credit card required.
What you can do with $1.00:
| Model | Approximate conversations |
|---|---|
deepseek/deepseek-v4-flash | ~200–300 short conversations |
anthropic/claude-sonnet-4 | ~25–40 short conversations |
openai/gpt-4.1-mini | ~100–150 short conversations |
Estimates assume ~500 input tokens + ~300 output tokens per turn, ~5 turns per conversation.
Model Pricing
All prices shown are UClaw prices for agentic runs (raw model cost + 15% infrastructure fee), per 1 million tokens. Plain text generation without a stateful agent is charged at raw model cost with no markup.
See the full Models page for the complete list with context window sizes.
Popular Models
| Model | Input / 1M tokens | Output / 1M tokens |
|---|---|---|
deepseek/deepseek-v4-flash | $0.16 | $0.32 |
deepseek/deepseek-v4-pro | $0.50 | $1.00 |
anthropic/claude-sonnet-4 | $3.45 | $17.25 |
anthropic/claude-opus-4.7 | $5.75 | $28.75 |
openai/gpt-4.1 | $2.30 | $9.20 |
openai/gpt-4.1-mini | $0.46 | $1.84 |
openai/gpt-5 | $1.44 | $11.50 |
google/gemini-2.5-flash | $0.35 | $2.88 |
google/gemini-2.5-pro | $1.44 | $11.50 |
Prices are subject to change as upstream providers adjust their rates.
Billing FAQ
When is my balance deducted?
Your balance is deducted in real time as tokens are generated. If a run is interrupted mid-stream, you only pay for the tokens already produced.
What happens when my balance runs out?
When your balance reaches zero, new agent runs will return a 402 Payment Required error with a clear message. Existing in-progress runs are not interrupted.
Is there a subscription or monthly fee?
No. UClaw is strictly pay-as-you-go. You are never charged unless you add funds manually.
Can I get a refund?
Unused balance is non-refundable. If you experience platform errors that charged your account (UClaw-side bugs, not model errors), contact support.
How do I monitor my spending?
The developer dashboard shows your real-time balance, usage history per API key, and per-run token breakdowns.