Skip to content
runlog_Guide
Menu
HomeEstimate cost
MODEL PRICING · BROWSER ESTIMATES

Estimate your AI costs.

See what one AI request could cost. Paste your prompt and choose how long you expect the answer to be. No account needed.

Your request (prompt)
Updates as you type

Example: draft a helpful reply to a customer.

See estimated cost ↓
0 tokens

Tokens are small pieces of text. The default 500-token answer is roughly a few paragraphs.

Expected answer length 500 tokens
Short (250) Structured (1,000) Code (4,000) Article / Context (16,000)

Estimated cost

Add a request or try an example to see an estimate.

Each price includes your prompt and assumed answer length for one call, unless you change usage below. This calculator does not send your prompt to a model. These are prices for software using AI services (APIs), separate from a ChatGPT subscription.

Model examples · total for 1 request

Text model example
Claude 3.7 Sonnet (Standard)
$0.00
Anthropic • 200k Context
Another text model example
Gemini 2.0 Flash
$0.00
Google • 1M Context
Image and voice price example
Generative Media & Voice Per Generation / Min
FLUX.1 Schnell (fal.ai)
$0.003
fal.ai • 4-Step Distilled Image Speed not measured
Token breakdown and prompt checks ⚡ Auto-Complexity: Simple (~0 thinking tokens)
Estimated Total Tokens
0
Input / Output Split
0 in / 500 out
System
User Prompt
Schema/Media
Thinking
Output
✓ Prompt Compact & Efficient
No trace dumps or redundant boilerplate detected. Optimal token efficiency.
Estimate more than one call
Estimate usage:
Add context, instructions or media
🔄 Context Compounding Simulator

In ChatGPT, Claude, and agent loops, this estimate assumes each follow-up sends the full prior conversation. Truncation, summarization, retrieval, server state and caching can change actual usage.

💬 Current Turn in Conversation Turn 1 (Fresh Chat)
Turn 1 (Fresh) Turn 5 (Brief) Turn 15 (Deep) Turn 40 (Bloated)
🤖 Avg Output per Past Turn (Response Length) 600 tokens
Short (150) Standard (600) Long / Code (1,500) Deep (3,000)
Accumulated Past Context Debt: +0 tokens
Agent guidelines, personas, or guardrails 0 tokens
OpenAI function definitions, Zod schemas, or Pydantic JSON 0 tokens

Media conversions are illustrative and model-dependent. A generic token price is not a verified media quote; check the provider’s supported modalities and billing units before spending.

High-res tiles (~765 tokens/img)
~1,500 tokens / minute
~650 tokens / page
~260 tokens / second
Multimodal token estimation 0 tokens
Reasoning token assumptions
Assumed billable reasoning tokens 0 tokens
Standard (0) Balanced (4,000) Deep Search (16,000) Max Reasoning (32,000)
Save estimate:
Heuristic estimate · output usage is assumed
Templates, price snapshot and developer utilities
Templates:
Prompt Pricer
Loading price snapshot…
Compare all model prices

Model price estimates

Estimates for the prompt and usage settings above. Check the price snapshot before budgeting.

Search or filter to narrow the list. Scroll the table to see all matching models and prices.

Model Cohort Context Input / 1M Output / 1M Thinking / 1M Total Cost
Use the CLI or explore other tools
PROMPT PRICER // FREE LOCAL CLI

Automate Prompt Cost Checks in CI/CD

Run local prompt cost estimates from your terminal. Download the versioned CLI and use it in your own development workflow at no cost.

✓ Versioned CLI download ✓ Local prompt estimates ✓ Included model catalog ✓ Source and checksums
Next / Public evidence

Explore measured model results

Public benchmarks show recorded responses to prepared tasks. Compare the method, sample counts, quality scores and measured costs with the assumptions in your estimate.

Explore public benchmarks →
Runtime feedback

Set local Loop-Guard boundaries

Add local retry detection and observed-usage checks around the agent you are designing. Use runtime signals to improve the next prompt and baseline.

Explore Loop-Guard →

Existing-user access.

Action copied to clipboard!