pricing

Simple credits for serious agent work.

Buy credits once. Use them across supported managed models. Track token usage by model, task, and day. 1 credit = $1.

Explorer

$20

20 credits

Individual trial after invite

Request invite
Buildermost popular

$100

105 credits (+5 bonus)

Active solo developer

Request invite
Team

$500

550 credits (+50 bonus)

Small team beta usage

Request invite
Scale

$2,000

2,300 credits (+300 bonus)

Heavy beta users

Request invite

Minimum purchase $20 · Paid credits expire after 12 months where legally permitted · Promotional credits expire after 90 days and are consumed first · Credits are not cash and are not transferable · Workspace compute is included during beta under fair-use limits.

model rates

Different models cost different amounts. We show the rate before you run.

Model familyBest forInput / 1MCached in / 1MOutput / 1MContext
DeepSeek V3Economy open modelsCheap simple edits, classification, summarization0.380.040.57128K context
Qwen3 CoderOpen coding modelsLow-cost coding tasks and experiments0.540.052.16256K context
Gemini 2.5 FlashGemini Flash / fast modelsLong-context scans and inexpensive iteration0.380.043.131M context
Claude Haiku 4.5Claude Haiku / small modelsQuick code questions and simple fixes1.250.136.25200K context
Claude Sonnet 5Claude SonnetDefault agentic coding tasks3.750.3818.751M context
GPT-5.1OpenAI GPT coding modelsCoding agents, refactors, tests1.560.1612.50400K context
Claude Opus 4.8Claude Opus / frontier modelsHard debugging, architecture, complex refactors6.250.6331.251M context

Rates are in credits per 1M tokens and include CloudAgent's platform margin · Last updated Jul 6, 2026 · Rate changes apply prospectively; running tasks keep the rate they started with.

estimate first

Know the cost before you delegate.

Estimate a task
Agentic work can involve multiple model calls. Estimates are not guarantees, but they help you avoid surprise usage.

# estimated cost

0.49credits

≈ $0.49 USD

5 iterations × (200K in + 40K out) · DeepSeek V3

spend controls

Credits stop runaway usage.

Hard task budget

Each task runs with a credit ceiling; the agent stops at the cap.

Daily user budget

Cap your own daily managed-model spend.

Low-balance warning

Warnings at 10 and 3 credits; new tasks blocked below 1.

Auto-stop

Running tasks pause or stop when their budget is reached.

Admin credit grants

Beta credits are granted manually and expire after 90 days.

Team budgets

Planned — shared budgets for team pilots.

billing faq

Questions about credits.