Grok 4.6 through the API: setup and pricing

Update, October 2026. Grok 4.6 is currently unavailable in Zukko. The current model list is on the Models page↗ or at GET https://api.zukko.pro/v1/models. Closest replacement: gpt-5.6-sol.

A separate account with the model provider is not always necessary. Through a gateway, models are available with the same key as the rest of the catalogue and connect by changing one field (the example below uses Grok 4.6's replacement, gpt-5.6-sol).

The request

curl https://api.zukko.pro/v1/chat/completions \
  -H "Authorization: Bearer fetch_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.6-sol",
    "messages": [{"role": "user", "content": "hello"}]
  }'

The format is OpenAI-compatible, so existing code needs no changes — a base URL of https://api.zukko.pro/v1 and the model identifier are enough.

Price and window

While Grok 4.6 was in the catalogue it cost $0.8 per million tokens on input and on output, with a one-million-token context window. The replacement, gpt-5.6-sol, costs $0.3 per million on input and output (the vendor's official price is $5/$30), also with a one-million-token window.

A flat rate means a long answer costs the same per token as a long prompt. For chat workloads, where output is comparable to input, that beats asymmetric pricing; for agent workloads, where input outnumbers output several times over, the input price still decides.

What to compare it on

Comparing models by public benchmarks is of little use: they measure an average over someone else's tasks, and models are deliberately tuned for the popular ones.

It is more useful to assemble your own set of five to seven tasks where you know the right answer, and run them across two or three candidates. Look not at how well the answers read, but at how many iterations it took to reach something usable and at the total token spend — which comes back in the usage field of every response.

Behaviour on long context deserves a separate check. A million tokens in the specification does not mean the model is equally attentive across the whole span; different models degrade differently, and it only shows on your own data.

What to watch financially

With a million-token window it is easy to run up a large bill without noticing: in agent workloads the context is resent at every step, and spend grows non-linearly.

A habit that helps: check usage during the run rather than only at the end, and clear history once it stops being useful. The costliest mistake is keeping already-read files in context until the session ends. Also keep in mind: vendors charge ~10% of the input price for prompt-cache reads, while Zukko bills cached tokens at the normal input price, so for agents with a large repeated context the real difference from the official price list is smaller than a head-on comparison suggests.

Switching

Every model in the catalogue works through one key and one balance, so trying a new one needs no new accounts. The current model list is at GET https://api.zukko.pro/v1/models, prices and context at GET /v1/models/info (or the Models page in the dashboard), and going back is a one-field edit.