Frontier labs · 8 November 2026
Available nowThinking Machines releases Inkling 2 Small multimodal MoE
Inkling 2 Small is a 594 billion‑parameter multimodal mixture‑of‑experts model with 17 billion active parameters, trained in 68 hours on 12 trillion tokens. It is available now on Tinker, with Apache 2.0 weights promised next week after external safety testing.
Benchmarks are the company's own; not independently verified.
Why it matters
Inkling 2 Small trained 594 B‑parameter model in 68 hours on 12 T tokens; benchmark your training pipelines against this speed to gauge competitiveness.
For product managers
For your product
What does this mean for your product?
Try it on your product More from this issueAll AI news from 8 Nov →
Get this every morning, written for your product.
Deltaday reads the whole AI news cycle and tells you what changed, whether it's real yet, and what to do about it.