How much does the Inkling API cost?

On FastMetal, Inkling is billed per token in yen: ¥178.7 per 1M input tokens and ¥723.74 per 1M output tokens, before tax. Usage is drawn from a prepaid balance; there is no subscription or monthly fee.

Can I call Inkling with the OpenAI SDK?

Yes. Point base_url at https://api.fastmetal.ai/v1 and pass "inkling" as the model; existing OpenAI-style code works unchanged, including streaming, tool calls and structured output.

What is the context window of Inkling?

524,288 tokens, shared between the prompt and the response.

How does Inkling rank?

#54 on the public arena's Japanese board (ELO 1,422). Ranks move as the leaderboard is updated.

What do I need to try Inkling?

Create an account and add credit; the model is then available both in the browser chat and over the API. There is no contract or minimum spend.

Back to Models

thinkingmachines/inkling

Inkling

Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total. It is designed for general-purpose reasoning, coding, agentic and tool-use systems,...

7/17/2026

524,288 tokens

#29 Text (Math)

Input: ¥178.7/M

Output: ¥723.735/M

Try in Chat Get API Key

Specifications

Modalities

Input

text

image

audio

Output

text

Supported Parameters

frequency_penalty

include_reasoning

logit_bias

max_tokens

min_p

presence_penalty

reasoning

reasoning_effort

repetition_penalty

stop

temperature

tool_choice

tools

top_k

top_p

Reasoning Configuration

Default: Thinking on (effort: high)
Selectable effort levels: max
high
medium
low
minimal
none
Turning thinking off: Possible (reasoning.enabled: false)

How to control thinking, and what it costs

Code Examples

curl https://api.fastmetal.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "inkling",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

How Inkling actually answers

Real responses to our standard prompts, recorded on FastMetal.

Count the number of 'r's in 'strawberry'

Count the number of 'r's in 'strawberry'. Explain your reasoning step by step.

Let's break down the word **"strawberry"** letter by letter: 1. **s** 2. **t** 3. **r** ← (1st 'r') 4. **a** 5. **w** 6. **b** 7. **e** 8. **r** ← (2nd 'r') 9. **r** ← (3rd 'r') 10. **y** So the spelling is: s-t-**r**-a-w-b-e-**r**-**r**-y Counting them up: - One 'r' in "str" (the 3rd letter) - Two 'r's in "berry" (the 8th and 9th letters) **Total: 3 'r's** in "strawberry".**

2.4s response417 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…

The error occurs because `users` is `undefined` when `UserList` renders, so `users.map` fails. This usually happens when: - The parent doesn't pass the prop (`<UserList />` instead of `<UserList users={...} />`) - Data is still loading (e.g., API hasn't returned yet) ## Quick Fix Provide a default empty array so `.map` always has an array to work with: ```jsx function UserList({ users = [] }) { return ( <ul> {users.map(user => <li key={user.id}>{user.name}</li>)} </ul> ); } ``` Alternatively, guard against undefined explicitly: ```jsx function UserList({ users }) { if (!users) return null; // or a loading state return ( <ul> {users.map(user => <li key={…

4.9s response585 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):…

Here is a review with an improved version followed by explanations. ### Issues in the original - **Imports inside the function**: Re-importing on every call is inefficient. - **Bare `except:`**: Catches `SystemExit`, `KeyboardInterrupt`, etc. - **Silent failures**: No logging; errors disappear. - **Magic numbers**: `10`, `2`, `200` are hard-coded. - **Data validation missing**: Assumes `data` is a list and every item has `"active"`, `"name"`, and `"id"` (will crash with `KeyError`). - **Performance**: Uses `.append()` in a loop instead of a list comprehension. - **Ambiguous return**: `[]` means both "success with no results" and "total failure". ### Improved version ```python import loggi…

54.1s response5649 output tokens

Compare these answers side by side with other models →

Frequently asked questions

How much does the Inkling API cost?: On FastMetal, Inkling is billed per token in yen: ¥178.7 per 1M input tokens and ¥723.74 per 1M output tokens, before tax. Usage is drawn from a prepaid balance; there is no subscription or monthly fee.
Can I call Inkling with the OpenAI SDK?: Yes. Point base_url at https://api.fastmetal.ai/v1 and pass "inkling" as the model; existing OpenAI-style code works unchanged, including streaming, tool calls and structured output.
What is the context window of Inkling?: 524,288 tokens, shared between the prompt and the response.
How does Inkling rank?: #54 on the public arena's Japanese board (ELO 1,422). Ranks move as the leaderboard is updated.
What do I need to try Inkling?: Create an account and add credit; the model is then available both in the browser chat and over the API. There is no contract or minimum spend.

Try Inkling right now

Inkling is available on FastMetal through one API key. Start in the browser, or call it from the OpenAI SDK.

Create a free account

Leaderboard

Text

OverallELO: 1,439

#83

JapaneseELO: 1,422

#54

ChineseELO: 1,481

#76

KoreanELO: 1,398

#66

EnglishELO: 1,446

#89

frenchELO: 1,451

#89

germanELO: 1,436

#80

spanishELO: 1,430

#83

russianELO: 1,426

#97

CodingELO: 1,491

#77

MathELO: 1,477

#29

Creative WritingELO: 1,380

#128

Instruction FollowingELO: 1,420

#98

Hard PromptsELO: 1,458

#90

Multi-TurnELO: 1,440

#94

Code

OverallELO: 1,409

#71

Compare with other models

See all comparisons →