Zukko Blog
Guides to Claude Code, Cursor, Codex and the Claude and GPT APIs: setup, token savings and troubleshooting.
Claude and GPT API for studying: how a student avoids overpaying
4 min · · бизнес
Five subscriptions for five different tasks makes no sense on a student budget. Here's how to get one key for every model and pay only for what you use.
A gateway for agencies and freelancers: per-client billing
5 min · · бизнес
When several client projects run through one account, the hard part isn't finding a model — it's not losing track of who spent what.
How to actually cut token spend
5 min · · экономия
Token spend rarely drops from one clever trick — usually it is five smaller decisions that add up to something noticeable.
What a very long context actually costs you
4 min · · payments
A million-token context window sounds like permission to stop worrying. In practice you pay for it in more than one currency.
Why input and output tokens are priced differently
4 min · · payments
Same token counter, two different rates. Here is where the gap comes from and why it matters for how you build.
How the price of a text request is calculated, start to finish
5 min · · payments
Between sending a request and paying for it there are several steps, and nearly every one is a place to either save money or lose track of spend.
How image generation pricing works
4 min · · изображения
An image is billed at a flat rate per piece, not per token. Where to find a model's exact price and how to confirm what a request actually charged.
Timeouts on long generations: how to avoid them
4 min · · development
A response cut off mid-stream almost always means your own client's timeout, not a token limit. Where it comes from and how to retry without a double charge.
Error 413: request too large — how to work around it
4 min · · errors
A 413 is almost never about a key or a model — it means the request body itself is too heavy. Where the extra weight comes from and how to cut it.
Image size and format limits: common mistakes
5 min · · изображения
An image error rarely names the exact cause outright. Where images actually get rejected on input, and how to debug it fast.
Reference-based image generation: how it works
4 min · · изображения
Reference mode carries a sample's style and composition into a new scene. The model, the request shape, and where it falls short.
Editing images through the API, step by step
4 min · · изображения
Editing is the same endpoint as generation, just a different model and an image in the request body. Walked through with real examples.
Full Node.js example: official SDKs through our gateway
2 min · · development
The Node builds of the official SDKs configure exactly like their Python counterparts — one baseURL field reroutes the whole request.
Full Python example: official SDKs through our gateway
3 min · · development
You don't need to replace the official SDKs, just repoint them. A working walkthrough of both clients against real endpoints.
No-code AI automation in Zapier and Make
4 min · · автоматизация
Zapier and Make don't ship a native block for our gateway, but the generic HTTP module covers everything a real integration needs.
OpenAI Agents SDK over our /v1/responses endpoint
4 min · · агенты
The Agents SDK talks to OpenAI by default. Two client settings redirect it through our gateway onto the models in our catalog.
LlamaIndex: indexing data with Claude and GPT
5 min · · development
LlamaIndex splits the answer-synthesis LLM from the embedding model — the gateway only covers the first part. Here's how to set that up honestly.
Vercel AI SDK: connecting a Next.js app to Claude and GPT
5 min · · development
The AI SDK's OpenAI and Anthropic providers build the path on top of baseURL differently — easy to get /v1 wrong in one route handler. Setup and the client-leak trap.
Semantic Kernel: connecting a .NET app to Claude and GPT
5 min · · агенты
SK talks to OpenAI through the official .NET SDK, which honors HttpClient.BaseAddress. Setup, the trailing-slash 404, and double-charged retries.
Microsoft AutoGen: pointing it at your own key instead of OpenAI
5 min · · агенты
AutoGen builds its client on top of openai-python, so base_url and api_key live right in config_list. Setup and the gotchas that show up first.
CrewAI: connecting agents to Claude and GPT with one key
5 min · · агенты
CrewAI defaults to OpenAI through LiteLLM, but the LLM class can point anywhere. Setup, per-agent model choice, and the errors that show up first.
Bulk request processing without hitting the rate limit
4 min · · development
One key with a spending limit on a batch job is a guaranteed 429 partway through. How to structure bulk processing so that doesn't happen.
Why a startup wants a gateway instead of billing each provider directly
5 min · · бизнес
Five providers means five bills, five currencies, five API shapes. What changes when one key and one balance sit behind all of it.
Neovim: connecting avante.nvim and codecompanion
4 min · · IDE
avante.nvim and codecompanion.nvim handle addresses differently under the hood — one provider expects /v1, the other doesn't. Both configs, line by line.