Inkling

Open weights, ready to tinker

Efficient open models for text, images, and audio, available to fine-tune.

Two sizes of Inkling

Our most capable open model, with native reasoning over text, images, and audio.

Parameters
975B total 41B active
Architecture
Mixture of Experts
Context
1M tokens (64K / 256K on Tinker)
Input
Text, Image, Audio

Matches its larger sibling on many benchmarks at a quarter of the size, with lower cost and latency.

Parameters
276B total 12B active
Architecture
Mixture of Experts
Context
1M tokens (64K / 256K on Tinker)
Input
Text, Image, Audio

Well-rounded, generalist models

Spider chart comparing Inkling, Nemotron 3 Ultra, GLM 5.2, GPT 5.6 Sol, and Claude Fable 5 on ten evaluations scored from zero to one hundred. Inkling is shown with the heavier cobalt line. Evaluations without a reported model score are plotted at zero. Hover an evaluation to compare every model's score.

The same capabilities, at every size

General intelligence

Performs well across knowledge, math, and science.

Agentic coding & tools

Writes code and uses tools to get work done.

Audio & image input

Natively multimodal on input, and at the open-source frontier for speech.

Controllable effort

Adjust thinking time to balance speed and performance.

Forecasting & confidence

Makes predictions with well-calibrated confidence.

Instruction following

Handles detailed, multi-step instructions reliably.

Ready to customize

Grab the open weights from Hugging Face, or fine-tune Inkling for your domain on Tinker, our training platform.