Pricing & Cost Optimization

Pricing Model

NeevCloud uses transparent, usage-based pricing that aligns costs with actual consumption. You're never paying for idle resources or overprovisioned capacity.

Pay-Per-Token Structure

How Token Billing Works

Tokens are the fundamental unit of text processing in language models. Roughly speaking:

• 1 token ≈ 4 characters of English text

• 1 token ≈ ¾ of a word

• 100 tokens ≈ 75 words

When you make a request:

• Your input (prompt, context, etc.) is counted as input tokens

• The model's response is counted as output tokens

• You're billed for both, with output tokens typically costing more

Example Calculation

Let's say a model charges:

• $0.50 per million input tokens

• $1.50 per million output tokens

If you send a 1,000-token prompt and receive a 500-token response:

• Input cost: (1,000 / 1,000,000) × $0.50 = $0.0005

• Output cost: (500 / 1,000,000) × $1.50 = $0.00075

• Total cost per request: $0.00125

For 10,000 similar requests per day, your monthly cost would be approximately $375.

Why Output Tokens Cost More

Generating text requires more computation than processing it. When you send input tokens, the model:

• Reads and encodes the text

• Builds internal representations

When the model generates output tokens, it:

• Runs complex neural network calculations for each token

• Samples from probability distributions

• Maintains coherence across the entire response

• Performs this iteratively for each output token

This is why output tokens typically cost 2-3x more than input tokens.

No Hidden Fees

What's Included in Token Pricing

Your per-token cost covers:

• GPU compute time for inference

• Model loading and memory allocation

• Load balancing and autoscaling

• Redundancy and failover

• Monitoring and logging infrastructure

• API gateway and networking

• Data transfer and bandwidth

You don't pay separately for any infrastructure components.

What You Don't Pay For

• Idle time between requests

• Model deployment or setup

• Minimum usage commitments

• Infrastructure maintenance

• Failed requests (you're only charged for successful requests that return tokens)

No Minimum Commitments

Complete Flexibility

You can:

• Start using a model with a single request

• Scale up to millions of requests per day

• Scale back down to zero requests

• Enable or disable models at any time

• Switch between models without penalties

There are no contracts, no reserved capacity requirements, and no charges when you're not making requests.

Ideal for Variable Workloads

This model works well when:

• Testing multiple models to find the best fit

• Running pilot programs with uncertain demand

• Building applications with unpredictable usage patterns

• Handling seasonal or event-driven traffic spikes

Real-Time Billing Visibility

Dashboard Transparency

Your NeevCloud dashboard shows:

• Current costs updating as requests complete

• Per-model cost breakdown

• Token usage trends over time

• Projected monthly costs based on current usage patterns

This visibility helps you:

• Stay within budget constraints

• Identify cost spikes immediately

• Make data-driven decisions about model selection and optimization

No Billing Surprises

You'll never receive an unexpected bill. The dashboard shows exactly what you'll be charged before the invoice is generated.

Cost-Efficient Scaling

Automatic Optimization

As your usage increases, the per-token efficiency improves because:

• Models stay loaded in memory (no repeated loading costs)

• Request batching optimizes GPU utilization

• Infrastructure is fully utilized, reducing per-request overhead

Right-Sizing Your Usage

Use the monitoring tools to optimize costs:

• If quality meets your needs with a smaller model, switch to save costs

• If response times are critical, upgrade to a larger model despite higher costs

• If output length is excessive, add max_tokens limits to control costs

The pay-per-token model gives you complete control over the cost-quality-performance tradeoff.