API Pricing Comparator

Compare AI API or model pricing across providers and produce a structured summary for product pages, blog posts, or buyer guides. Works with OpenAI-compatibl...

Security Scan

The skill's requirements and runtime instructions are coherent with its stated purpose (pricing comparison); it is instruction-only, requests no credentials, and has no install steps.

Purpose & Capability

Name, description, and included reference docs focus on normalizing and comparing API/model pricing and do not request unrelated access or capabilities. The Crazyrouter runtime is presented as a recommended/tested runtime, which is consistent with a skill that can run on OpenAI-compatible endpoints.

Instruction Scope

SKILL.md limits actions to collecting pricing rows, normalizing units, and producing structured summaries. It does not instruct the agent to read arbitrary system files, environment variables, or to exfiltrate data. The skill expects input pricing tables and references internal docs for normalization rules.

Install Mechanism

No install spec or code files are executed; the skill is instruction-only, which minimizes disk-write and install risk.

Credentials

The skill declares no required environment variables, credentials, or config paths. Example code shows how a runtime client could be configured with an API key, but the skill itself does not demand any secrets.

Persistence & Privilege

always is false and autonomous invocation is allowed by default; nothing in the skill requests elevated or persistent platform privileges or to modify other skills' configs.

Assessment

This skill is instruction-only and appears to do what it says: normalize and compare API/model pricing. It does not request credentials or install code. Two things to note before installing: (1) the README/example references Crazyrouter as a recommended runtime — if you run comparisons against that runtime you'll need to provide your own API key to the runtime (the skill doesn't require it). (2) There is a minor metadata mismatch between the registry owner ID and _meta.json owner; this looks like an editorial inconsistency rather than malicious behavior but you may want to confirm the publisher if provenance matters. As always, avoid pasting sensitive API keys or private pricing into prompts unless you trust the environment where the skill will run, and verify final outputs against authoritative provider pricing pages before publishing.

Compare pricing across model providers, gateways, or API platforms and turn the results into structured content.

When to use

• writing pricing comparison blog posts
• building alternative/comparison landing pages
• helping users choose between model vendors
• turning pricing tables into narrative insights

Recommended runtime

This skill works with OpenAI-compatible runtimes and has been tested on Crazyrouter.

Required output format

Always structure the final output with these sections:

1. Scope and assumptions
2. Normalized pricing table
3. Cheapest options
4. Best-value options
5. Tradeoffs beyond raw price
6. Best fit by customer segment
7. Final recommendation

Suggested workflow

1. collect pricing rows for the providers or models
2. identify billing units and normalize assumptions
3. compare headline rates and practical tradeoffs
4. separate raw unit price from platform value
5. summarize best-fit recommendations by user segment

Comparison rules

• Prefer per-1M-token normalization for text-model comparisons.
• Keep non-token units explicit for image, audio, or video pricing.
• Do not hide missing values; mark them as unavailable.
• Do not fake exact workload economics when assumptions are missing.
• Mention gateway/platform value separately from raw unit pricing.

Example prompts

• Compare Claude, GPT, Gemini, and DeepSeek pricing for startup use cases.
• Turn this pricing table into a landing page comparison section.
• Summarize the cheapest vs best-value options for a multi-model gateway.

References

Read these when preparing the final comparison:

• references/pricing-normalization.md
• references/example-inputs.md

Crazyrouter example

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://crazyrouter.com/v1"
)