Concepts

Core ideas that influence latency, quality, and cost when integrating Nascentist.

1. Models

nascentist-1 is optimized for coding workflows including code generation, debugging, and review. You do not need multiple model families for most use cases.

Model	Context Window	Max Output	Status	Best For
nascentist-1	8,192 tokens	4,096 tokens	Available	Code generation, debugging, review

2. Tokens

Tokens are the billing and context unit. A rough rule of thumb is ~4 characters ≈ 1 token, but code tends to consume more tokens than plain prose because of symbols, indentation, and identifiers.

4 lines

def total_cost(items):
    subtotal = sum(item["price"] for item in items)
    tax = subtotal * 0.18
    return round(subtotal + tax, 2)

Note

Use the token counter in the Playground to estimate costs before integrating.

3. Completions vs Agentic

Complete is one-shot generation: prompt in, output out. It is cheaper and lower latency for straightforward tasks.

Agent Fix is iterative: generate, execute, observe errors, revise, and repeat. It costs more than one-shot completion but is better for unresolved runtime issues.

Mode	Flow	Latency	Cost
Completion	Predict → Return text	Lower	Lower
Agent Fix	Predict → Execute → Observe → Fix	Higher	Higher

4. Rate Limits

Limits are enforced per plan and can vary by request, minute, and daily windows. See the full breakdown in Rate Limits.