Thinking MachinesBase plan

Inkling 2 Small

Thinking Machines Lab's second-generation Inkling via the Vercel AI Gateway: an open-weights multimodal MoE (594B total, 17B active) with text and image understanding, controllable thinking effort, and function calling at a 1M-token context window. Built for coding, tool use, and professional knowledge work. Uses 2 base requests per send before length multipliers. Supports image input. Web search runs through Perplexity.

The model you pick is the model that answers. Some replies follow our crisis protocol and may be fixed text.

About Inkling 2 Small

Inkling 2 Small is a Thinking Machines model. It has a context window of 1,048,576 tokens and can return up to 262,144 tokens per response. It accepts text and image input and returns text.

Suggested uses

  • Coding and tool-using assistants with adjustable thinking effort
  • Long-document and image analysis on a small-model budget

Specifications

ProviderThinking Machines
Released2026-11
IntelligenceHigh
SpeedFast
Context window1,048,576 tokens
Max output262,144 tokens
Knowledge cutoffNot officially published
Input price$0.30 / 1M tokens
Output price$1.00 / 1M tokens
Request cost2 base requests
Plan tierBase
InputText, Image
OutputText
FeaturesCached input: $0.03 / 1M tokens, 2 base requests per send before length multipliers, Open weights, Controllable thinking effort, Function calling supported, Image input supported, Served through the Vercel AI Gateway (thinkingmachines/inkling-2-small), Zero data retention and no prompt training
Model IDinkling-2-small

Specifications and list prices are as published by the model's provider and can change. Request cost is what the model uses on just4o.chat.

Frequently asked questions

The provider list price is $0.30 / 1M tokens for input and $1.00 / 1M tokens for output. On just4o.chat, Inkling 2 Small is a Base model and uses 2 base requests per send before length multipliers.

Input: text, image. Output: text.

Inkling 2 Small is the second-generation model with 17B active parameters (594B total), a 1M-token context window and a lower price. The original Inkling is larger (41B active) and also understands audio.

Related models