Models·Thinking Machines

Inkling: Open-Weight Model Pricing and Limits

thinkingmachines/inkling

Inkling is an open-weight model for reasoning, coding, images, audio, and tool-use systems.

$0.95Input, per million tokens
$4.05Output, per million tokens
524KContext window
262KLongest answer

524K tokens of context — about 393,216 words, on a scale that ends at 2M.

What it can do

  • Reasoning
  • Vision
  • Tool use
  • Open weights
  • Takes text
  • Takes image
  • Takes audio
  • Gives text

Released 17 Jul 2026 · checked 3 Oct 2026.

About it

Inkling is a Thinking Machines Lab model with open weights and 41 billion active parameters from a 975 billion-parameter mixture-of-experts design. It accepts text, images, and audio, supports reasoning and tools, and returns text. Its context window is 524,288 tokens, with up to 262,144 output tokens. Pricing is $0.95 per million input tokens and $4.05 per million output tokens.

Review

In a line

Inkling is an open-weight multimodal model from Thinking Machines Lab. It handles text, images, and audio, then produces text answers.

What it's good at

The model is built for general-purpose reasoning, coding, and agentic systems that use tools. Reasoning and tool support are included, and its multimodal input lets it work with images and audio as well as text.

Inkling uses a mixture-of-experts design with 41 billion active parameters out of 975 billion total. Its weights are open, so it belongs to the open-weight model group.

Price in plain terms

Input costs $0.95 per million tokens, and output costs $4.05 per million tokens. Tokens are small pieces of text or other input counted for billing.

Reading a page of about 500 words costs about 0.062 cents. A chat turn with about 1,000 input tokens and 500 output tokens costs about 0.3 cents. At that same size, 1,000 turns cost about $2.97.

Context and limits

Inkling can take up to 524,288 tokens in one context, so a single request can include a large amount of material. It can produce at most 262,144 tokens in one answer.

The model's listed output is text.

Ways to use it

Inkling is intended for general-purpose apps, coding tools, agentic systems, and other software that can call tools. Its open weights and model identifier, thinkingmachines/Inkling, provide a way to identify the model when setting it up.

It is one of the Thinking Machines models and can be considered alongside the newest models.

Verdict

Choose Inkling if you need one model that can reason, work with text, images, and audio, and take part in tool-using systems. Its very large context and output limits suit tasks that keep substantial material together.

Skip it if you need a non-text output format or if your software cannot work with open model weights.

Verdict

Inkling fits teams building software around reasoning, coding, and tool calls, especially when image or audio input matters. Open weights give builders more control over deployment choices, while the large context can hold long inputs. Costs remain low for short chats but rise with generated output, which is priced higher than input.

  • Accepts text, images, and audio inputs
  • Supports reasoning and tool-use systems
  • Open weights are available
  • 524,288-token context window
  • Up to 262,144 output tokens
  • Returns text only
  • No ready-made integrations are listed
  • Output tokens cost more than input tokens
  • Large model has 975 billion total parameters
  • Pick it ifReasoning and coding applications; Agents that use tools; Apps handling text, images, and audio
  • Skip it ifYou need image, audio, or other non-text output; You require listed plug-and-play integrations; Your software cannot use open model weights

Pricing

You pay $0.95 per million input tokens and $4.05 per million output tokens. Short chats cost about 0.3 cents each at the stated example size; 1,000 such turns cost about $2.97.

Questions

What is Inkling?

Inkling is an open-weight multimodal model from Thinking Machines Lab for reasoning, coding, agentic work, and tool use.

What can Inkling take as input?

It accepts text, images, and audio. Its listed output is text.

How much does Inkling cost?

Input is $0.95 per million tokens. Output is $4.05 per million tokens.

How large is Inkling's context window?

It can handle up to 524,288 tokens in one context.

How long can an Inkling answer be?

The maximum output is 262,144 tokens.

What does an example chat cost?

A turn with about 1,000 input tokens and 500 output tokens costs about 0.3 cents. 1,000 such turns cost about $2.97.

Are Inkling's weights open?

Yes. The model is listed with open weights, under the identifier thinkingmachines/Inkling.

On these boards

More from Thinking Machines