Quick Start
- 1
- 2Use the Tinker SDK or REST API to call
thinkingmachines/Inkling— see the documentation for request format. - 3
Code Examples
About Tinker
Tinker exposes low-level training and sampling primitives while managing the underlying infrastructure. Inkling is available in 64K and 256K context configurations, and the platform publishes token-based prefill, sample, and training prices.
Tinker is Thinking Machines Lab's managed API platform for fine-tuning and sampling supported foundation models. It is the first-party public Tinker surface through which Inkling is available for fine-tuning.
Pricing on Tinker
| Type | Price (per 1M) |
|---|---|
| Input tokens | $1.87 |
| Output tokens | $4.68 |
Capabilities
VisionMultimodalReasoningTool UseAudioFine-tuning
About Inkling
Inkling is Thinking Machines Lab's Apache-2.0 open-weight general-purpose multimodal model. It accepts text, image, and audio inputs and generates text with a 1M-token context window. Compare it for Coding, RAG, Agents, Long context, Vision, and JSON / Tool use.
Model Specs
Released2026-07-15
Parameters975B total, 41B active
Context1m
ArchitectureMixture of Experts