Pricing
Simple, usage-based pricing for models and compute.
Models
Section titled “Models”Pay per token. The same rates apply across the Responses API, Chat Completions, and Tasks.
| Model | Purpose | Input | Output |
|---|---|---|---|
tzafon.northstar-cua-fast-1.6 | Our latest computer-use model. Recommended; pin this version in production. | $0.50/M tokens | $1.50/M tokens |
tzafon.northstar-cua-fast | Unversioned alias of the previous generation. Prefer the pinned -1.6. | $0.50/M tokens | $1.50/M tokens |
tzafon.sm-1 | Text model specialized for low-level systems programming. | $0.20/M tokens | $0.30/M tokens |
List the models available to your key at any time with client.models.list().
CUA token overhead
Section titled “CUA token overhead”Each request using the computer_use tool (via the Responses API) incurs approximately 1,974 additional input tokens of fixed overhead.
Compute
Section titled “Compute”Pay only for what you use, billed per second.
| Resource | Rate |
|---|---|
| CPU | $0.38 / core-hour |
| Memory | $0.095 / GiB-hour |
A normal browser workload runs at roughly $0.075 / hour.
How billing works
Section titled “How billing works”- CPU. Based on actual core usage (2 cores at 100% = 2 core-seconds per second)
- Memory. Based on actual memory used, not allocated
- Granularity. Per-second billing, minimum 0.125 cores per container
What does a task cost?
Section titled “What does a task cost?”A task’s cost has three drivers, all visible in the rates above:
- Input tokens per step. Each computer-use step sends the fixed ~1,974-token tool overhead plus the screenshot, and conversation context accumulates as the task runs.
- Output tokens. Small; actions are short (tens of tokens per step).
- Compute time. The browser or desktop’s runtime at roughly $0.075/hour for a typical browser workload.
In practice, a typical 10–20-step task lands between a few cents and ~$0.50, driven mostly by how much context accumulates over the run. Two levers keep costs predictable: max_steps caps the worst case, and max_duration_seconds caps compute. Watch real per-task numbers in the usage dashboard. The Tokens/Infra/Spend views break down exactly where each run’s cost went.
What if I forget to terminate? Instances automatically terminate after a configurable inactivity period to prevent unexpected charges.
Are there setup fees? No setup fees, no monthly minimums. Pay only for what you use.
Volume discounts? Contact support@tzafon.ai for high-volume pricing.
Next steps
Section titled “Next steps”- Quickstart: run your first task in under 5 minutes
- Running in production: control costs with
max_steps, timeouts, and session hygiene - How Lightcone works: models and the API layers
- Usage dashboard: live token, compute, and spend breakdowns