Inkling 2 Small
Thinking Machines Lab’s second-generation small model. Launched on Tinker on Nov 5, 2026; open weights (Apache-2.0) on Hugging Face from Nov 12.
PricePrice not published
Checked13 November 2026
No published training cutoff
At least110events since its release on 5 November 2026
3 historic, 27 major. 8 days of news.
Thinking Machines Lab doesn’t publish a training cutoff, so we count from its release date. The real gap is larger.
How to call it
3 ways to use Inkling 2 Small.
| Provider | Model id | Endpoint | Docs |
|---|---|---|---|
| Tinker (Thinking Machines) | thinkingmachines.ai | ||
| Hugging Face | huggingface.co | ||
| Hugging Face (MXFP8) | huggingface.co |
Pricing
Price not published.
No price source linked. Checked 13 November 2026.
Specs
- Input
- Text and images
- Output
- Text
- Context window
- 262,144 tokens
- Max output
- Not published
- Open weights
- Yes
- Licence
- Apache-2.0
- Training cutoff
- Not published
- Released
- 5 November 2026
- Status
- Current
- Type
- Reasoning model
What stands out
What its maker marketed at launch, and what people found later, each with its source.
Strong ARC-AGI score for an open-weights model
First
ARC Prize verified ARC-AGI-2 81.7% at $0.27/task and ARC-AGI-1 96.5% at $0.08/task, above every open-weight model ARC had evaluated (Nov 6, 2026).
Notes
Weights published Nov 12, 2026 after external safety testing; Artificial Analysis Intelligence Index about 31 (v5).
Sources: https://thinkingmachines.ai/news/inkling-2-small/ · https://huggingface.co/thinkingmachines/Inkling-2-Small · https://x.com/miramurati/status/2120983826407268458
Not yet checked against an official source: 594B total / 17B active MoE with a 292M-parameter vision encoder; trained with 256K context plus compaction (context_window here is that training length; served limits not verified). Pricing on Tinker not verified.