GPT-5.6 Terra vs Kimi K3
Pricing, context window and real answers (August 2026)
GPT-5.6 Terra (OpenAI) and Kimi K3 (Moonshot), compared as you can actually call them on FastMetal. Both are served from the same OpenAI-compatible endpoint and API key; switching is a change to the model string.
Specs and pricing
 GPT-5.6 Terra |  Kimi K3 | |
|---|---|---|
| Provider | OpenAI | Moonshot |
| Input (per 1M tokens) | ¥357.4 | ¥536.1 |
| Output (per 1M tokens) | ¥2,144.4 | ¥2,680.5 |
| Typical cost (1,000 in + 500 out tokens × 1,000 calls) | ¥1,430 | ¥1,876 |
| Context window | 1,050,000 tokens | 1,048,576 tokens |
| Release date | 7/9/2026 | 7/16/2026 |
| Input modalities | file, image, text | text, image |
| Arena · overall | #34 · ELO 1,468 | #11 · ELO 1,486 |
| Arena · Japanese | Unranked | Unranked |
| Arena · coding | #22 · ELO 1,521 | #8 · ELO 1,530 |
Prices are FastMetal's yen rates before tax. Arena ranks are as of the latest leaderboard scrape.
Which should you pick?
- On price, GPT-5.6 Terra: ¥2,144.4 per 1M output tokens, 20% less than Kimi K3.
- On the Overall arena board, Kimi K3 ranks higher (#11; GPT-5.6 Terra is #34).
- On the Coding arena board, Kimi K3 ranks higher (#8; GPT-5.6 Terra is #22).
- For long documents, GPT-5.6 Terra: a 1,050,000-token context window against 1,048,576 for Kimi K3.
- If in doubt, try both on the same key. Switching is a change to the model string, and each is billed at its own rate.
Real answers to the same prompts
Exactly what the FastMetal gateway returned, side by side. Not benchmark scores: actual output.
Count the number of 'r's in 'strawberry'
Count the number of 'r's in 'strawberry'. Explain your reasoning step by step.
GPT-5.6 Terra
“strawberry” is spelled:
s t **r** a w b e **r** **r** y
There are **3** occurrences of the letter **r**.
3.6s71 output tokens
Kimi K3
# Counting the 'r's in "strawberry"
**Step 1: Write out each letter with its position**
| Position | Letter |
|----------|--------|
| 1 | s |
| 2 | t |
| 3 | **r** ✓ |
| 4 | a |
| 5 | w |
| 6 | b |
| 7 | e |
| 8 | **r** ✓ |
| 9 | **r** ✓ |
| 10 | y |
**Step 2: Identify each 'r'**
- Position 3: the 'r' in "st**r**"
- Position 8: the first 'r' in "be**r**ry"
- Position 9: the second 'r' in "ber**r**y"
**Step 3: Verify by splitting the word**
"Strawberry" = "straw" + "berry"
- "straw" contains **1** r
- "berry" contains **2** r's
**Answer: There are 3 r's in "strawberry."** 🍓
12.2s594 output tokens
Debug This Error
I'm getting the following error in my Node.js application: TypeError: Cannot read properties of undefined (reading 'map') at UserList (/app/components/UserList.js:12:25) at renderWithHooks (/app/node_modules/rea…
GPT-5.6 Terra
The error means `users` is `undefined` when `UserList` renders, so React tries to evaluate:
```js
users.map(...)
```
But `.map()` only exists on arrays.
Common causes:
- The parent component does not pass a `users` prop.
- The user data is loaded asynchronously and has not arrived on the first render.
- The API response shape differs from what you expect, e.g. data is in `response.data.users` rather than `response.data`.
- The parent initializes its user state as `undefined` or `null`.
## Fix options
### 1. Initialize users as an empty array
In the parent component:
```jsx
const [users, setUsers] = useState([]);
```
Then pass it normally:
```jsx
<UserList users={users} />
```
An e…
5.8s559 output tokens
Kimi K3
## What's causing it
The error means `users` is `undefined` when the component renders — `.map()` only exists on arrays, so calling it on `undefined` throws. Your component code is fine; the problem is **upstream**: whatever renders `<UserList />` isn't passing a valid array.
The most common culprit is **async data fetching with uninitialized state** in the parent:
```jsx
function App() {
const [users, setUsers] = useState(); // ← undefined!
useEffect(() => {
fetch('/api/users')
.then(res => res.json())
.then(setUsers);
}, []);
return <UserList users={users} />; // undefined on first render
}
```
The component renders *before* the fetch resolves, so `users` is `…
47.5s1756 output tokens
Code Review
Please review the following Python function and suggest improvements for readability, performance, and best practices: def get_data(url, retries=3): import requests import time for i in range(retries):…
GPT-5.6 Terra
Here are the main areas to improve:
### Issues in the original code
- **Imports inside the function**: Move `requests` and `time` imports to module scope.
- **Bare `except:`**: This catches everything, including `KeyboardInterrupt` and programming errors. Catch `requests.RequestException` (and possibly JSON decoding errors) instead.
- **Status code check**: Prefer `response.raise_for_status()` or `response.ok` over manually checking only for `200`. Successful responses can include other 2xx status codes.
- **`== True`**: Use `if item.get("active"):` instead.
- **Potential `KeyError`**: `item["name"]`, `item["id"]`, and `item["active"]` can fail if the API response is incomplete. Use `.get(…
10.6s939 output tokens
Kimi K3
This works, but there are several issues worth addressing. Here's a breakdown, followed by an improved version.
## Issues
### Correctness & best practices
1. **Bare `except:`** — catches *everything*, including `KeyboardInterrupt` and programming bugs like `KeyError` on a malformed item (which then get pointlessly retried). Catch `requests.RequestException` specifically.
2. **Silent failures** — errors are swallowed with no logging, and returning `[]` makes "API is down" indistinguishable from "no active items." At minimum, log; ideally, raise after the final attempt.
3. **Retrying non-retryable errors** — a 400/401/404 will never succeed on retry, but the code sleeps and retries anyway. O…
57.6s8060 output tokens
Frequently asked questions
- Which is cheaper, GPT-5.6 Terra or Kimi K3?
- Per 1M output tokens, GPT-5.6 Terra is ¥2,144.4 and Kimi K3 is ¥2,680.5 on FastMetal (yen, before tax), so GPT-5.6 Terra is cheaper.
- How do the context windows of GPT-5.6 Terra and Kimi K3 compare?
- GPT-5.6 Terra takes 1,050,000 tokens; Kimi K3 takes 1,048,576.
- Which ranks higher, GPT-5.6 Terra or Kimi K3?
- Kimi K3 ranks higher on the public arena (GPT-5.6 Terra #34, Kimi K3 #11). Ranks move as the leaderboard updates.
- Can I use GPT-5.6 Terra and Kimi K3 with the same API key?
- Yes. On FastMetal's OpenAI-compatible endpoint, switch by passing "gpt-5.6-terra" or "kimi-k3" as the model. Each is billed at its own rate from the same prepaid balance.
Try both on one API key
Create an account and add credit to call GPT-5.6 Terra and Kimi K3 from the browser chat and the API. No monthly fee.