Thinking Machines has unveiled the Inkling 2 Mini, a state-of-the-art looped Mixture of Experts (MoE) transformer boasting 26.4 billion parameters. This innovative model features 3.4 billion active parameters and is designed to run seamlessly on a 32GB Mac or a DGX Spark system.

The Inkling 2 Mini has achieved an impressive score of 79.2% on the SWE-bench Verified benchmark, showcasing its capabilities in handling complex tasks. This launch positions Thinking Machines as a competitive player in the AI landscape, particularly for users seeking powerful models that can operate on accessible hardware.


Compiled automatically by the Tech AI Newsdesk from public AI-news sources and summarised in our own words.

Frequently asked questions

What is the Inkling 2 Mini?

The Inkling 2 Mini is a state-of-the-art looped Mixture of Experts (MoE) transformer launched by Thinking Machines.

How many parameters does the Inkling 2 Mini have?

The Inkling 2 Mini has a total of 26.4 billion parameters, with 3.4 billion active parameters.

What benchmark score did the Inkling 2 Mini achieve?

The Inkling 2 Mini achieved an impressive score of 79.2% on the SWE-bench Verified benchmark.