Cost Comparison
Plans, credits, and runtime cost
Direct LLMs meter tokens on every request. Sparkient uses a compiled runtime model and meters decisions, training, generation, labelling, and hosting through plan credits.
Plan
Price
Included credits
Decision types
Best fit
Trial
$0
5,000 once
1
Validate one decision
Developer
$19/mo
10,000/mo
2
Hobby and early projects
Starter
$199/mo
50,000/mo
3
Production projects
Growth
$599/mo
200,000/mo
10
Scaling and edge export
Scale
$1,999/mo
1,000,000/mo
50
High-volume systems
A deployed compiled decision usually uses about one credit. Training uses 2,000 credits. Generation, labelling, model serving, and LLM escalation also use credits.
Compare the two approaches
Hosted LLM API
Sparkient
Runtime
Live provider model
Compiled model
Billing
Tokens or requests
Credits by operation
Latency
Depends on the model and prompt
Under-100ms compiled-path target
Customisation
Prompting or fine-tuning
Trained on your examples
Offline use
Hosted provider connection required
Edge bundle on eligible plans
Budget
Varies with usage
Included credits and optional top-ups
Sparkient's 33–42ms figures are batch-average time per item in four controlled synthetic runs, not per-request p95. LLM escalation is model dependent.
Start Free →
5,000 one-time credits. No credit card required.