Developer tools that actually work in your browser

• LLM Tools OpenAI Token Counter Count tokens for GPT-4o, GPT-4, GPT-3.5, o1, o3 with cost estimates.
• LLM Tools Claude Token Counter Estimate tokens and cost for Claude Opus 5.5, Sonnet 5.5, Opus 4.7, Sonnet 4.6 and Haiku 4.5.
• LLM Tools Gemini Token Counter Count tokens for Gemini 1.5 Pro, Flash, and Gemini 2.x.
• Crypto & Hashing JWT Decoder Decode, verify, and generate JWTs. HS256, RS256, ES256, EdDSA support.
• JSON JSON Formatter & Validator Format, validate, and minify JSON. Strict and loose parsing modes.
• JSON JSON to TypeScript Generate TypeScript interfaces from JSON with deep nested type inference.
• Code Conversion cURL Converter Convert cURL commands to Python (requests/httpx), JavaScript (fetch), Go, and PHP.
• Regex Regex Tester Test regular expressions against text. PCRE, JavaScript, and Python flavors.
• LLM Tools LLM API Cost Calculator Estimate per-call and monthly costs across LLM providers.

Tested explanations for the edge cases behind the tools, including token sizing, JSON validation, JWT verification, and regex performance.

• Regex Regex Lookahead and Lookbehind, Explained With Practical Examples Positive and negative lookahead and lookbehind, explained with runnable password and delimiter examples, plus which regex engines actually support them.
• Security Where to Store JWTs in the Browser: localStorage vs Cookies vs Memory (XSS and CSRF) localStorage, cookies, or memory for JWTs? The real XSS-vs-CSRF tradeoff behind the in-memory access token plus HttpOnly refresh cookie pattern.
• Security How to Verify a JWT Signature: HMAC, RSA/ECDSA Public Keys, and JWKS/kid Verify a JWT signature correctly: HS256 shared secrets vs RS256/ES256 public keys, how kid picks a key from JWKS, and why kid, jku, and x5u are hints not contro
• Security JWT Structure Explained: Header, Payload, and Signature (With a Decoded Example) A byte-level walkthrough of a real JWT: header, payload, signature. Why Base64url is encoded, not encrypted, and why decoding a token is not verifying it.
• JSON Fix Invalid JSON: How to Repair Trailing Commas, Single Quotes, and Broken LLM Output Repair invalid JSON from LLMs, JS object literals, and configs: trailing commas, single quotes, unquoted keys, and truncated output. A practical, honest guide.
• JSON How to Convert CSV to JSON Without Breaking on Quoted Commas Convert CSV to JSON the right way: quoted commas, escaped quotes, the BOM, type inference, and where nesting is post-processing, not a converter feature.

Why these tools, why this site

Most "online tools" sites either bury the actual tool under three pop-ups and an ad wall, or they ship a single-purpose page that's been copy-pasted across fifty domains. Both patterns waste your time. Codeswap is built on a different default: the tool you came for loads in under a second, runs entirely in your browser, and never sends what you paste anywhere. The supporting content exists for context — not to bury the tool you came for.

Every page on this site is opinionated where it can be. The OpenAI Token Counter uses OpenAI's tokenizer through the js-tiktoken library (with an estimate as fallback if it can't load), not a regex approximation. The JWT Decoder verifies signatures across HS256, RS256, ES256, and EdDSA — and warns you when a token uses alg:none or an expired claim. The LLM API pricing table lists input, output, cached-input and batch prices for current Anthropic, OpenAI and Google models, each row copied from the provider's pricing page with the date it was checked.

The site is funded by Google AdSense (when approved) and the long-term plan is to keep the tools themselves free. If a tool becomes substantial enough to warrant a paid tier, the free version stays functional. See the About page for the operator's name and contact details.

Prices per million tokens for current models, each with its source and the date it was checked. Use it with the workload cost guide to estimate per-call cost, or compare two Claude models on cost at /compare/. Confirm on the provider's page before you budget.

• GPT-5OpenAI — $1.25/M input
• Claude Sonnet 4.6Anthropic — $3/M input, 1M context
• Gemini 2.5 ProGoogle (deprecated) — $1.25/M input
• Claude Opus 5.5Anthropic — $4/M input
• Claude Sonnet 5.5Anthropic — $2/M input
• GPT-5 miniOpenAI — $0.25/M input

How the tools work

Every tool on Codeswap is a static HTML page with a JavaScript front-end. When you load one, you get the full interface and content immediately. When you click "run," the computation happens in your browser — token counters use the real WebAssembly tokenizer for each provider, the JWT decoder uses the WebCrypto API for signature verification, JSON tools use the browser's native parser.

Nothing about your input is sent to a server. You can verify this in your browser's DevTools network tab. This matters because developers paste sensitive things into these tools — API responses, production JWTs, customer PII inside JSON dumps. A tool that leaks any of those is worse than no tool at all.

The pricing table, cost comparisons and workload cost guide are built from one price file copied from each provider's own pricing page, with the date it was checked. Prices change; if you spot one that is out of date, email [email protected] and it will be corrected.