Reported News
Thinking Machines Inkling 2 Small: 594B MoE, AA Index 31, closed weights on Tinker
Thinking Machines Lab introduced Inkling 2 Small on November 5, 2026: a Mixture-of-Experts model with 594B total parameters and 17B active, image inputs, and a score of 31 on the Artificial Analysis Intelligence Index. It is available on Tinker with closed weights; the lab says it intends Apache 2.0 weights next week after safety testing.
- Changes reported
- 6
- listed below, each from a source we read
- We tested
- Not tested yet
- reported from sources; we did not run it ourselves
- Gaps we flag
- 3
- things we could not verify
- Sources
- 1
- 1 primary, linked below
What changed
| # | What the source says | Status |
|---|---|---|
| 01 | New second-generation model: Inkling 2 Small is a Mixture-of-Experts model with 594B total parameters, 17B active, image inputs and variable effort. | Reported |
| 02 | Thinking Machines reports a score of 31 on the Artificial Analysis Intelligence Index, above its previous Inkling models, with gains claimed for agentic coding, vision, knowledge work and reasoning. | Reported |
| 03 | Initially available through Tinker with closed weights for playground use, fine-tuning and sampling; checkpoints are downloadable when weights ship. | Reported |
| 04 | The lab says it intends to release full weights under an Apache 2.0 license next week after external safety testing and red-teaming. | Reported |
| 05 | Training details in the post include an 8.4T-token H1 pretrain on GB300 NVL72, a joint RL run of 1,075 steps and over 17 million rollouts, and a 292M-parameter vision encoder. | Reported |
| 06 | Vendor-reported coding scores include 25.4% on FrontierCode v2.0, 72.1% on DeepSWE v1.1 and 86.3% on SWEBench Pro v2 (Hard); vision claims include 64.7% on PerceptionBench. | Reported |
reported by the sourcemeasured or run by usnot verified
Primary source: Thinking Machines Lab: Introducing Inkling 2 Small. We are the fast, independent index; the vendor’s own page is the source of record for this release.
In the source’s words
“It is a Mixture-of-Experts model with 594B total parameters, 17B active, and supports image inputs and variable effort.”
“Inkling 2 Small far surpasses our previous models' scores on the Artificial Analysis Intelligence Index with a score of 31”
“We intend to make the full weights available under an Apache 2.0 license next week after the completion of external safety testing and red-teaming.”
Early verdict
A major Thinking Machines scoop for agentic coding and vision if the AA Index and Tinker availability hold; treat open weights as pending until the promised Apache 2.0 drop.
Basis: Based on the lab's own November 5 announcement; we have not tested the model.
What we have not verified
- We did not run Inkling 2 Small on Tinker or reproduce any benchmark scores.
- Apache 2.0 weight release is stated as an intention after safety testing; timing and final license text are not confirmed.
- The detailed benchmark tables are vendor-reported; we did not cross-check third-party leaderboards for this brief.
This brief is Reported: written from the sources below, not from running the release ourselves.
Sources
- Thinking Machines Lab: Introducing Inkling 2 Smallprimary source · fetched 2026-11-06
Related
Spotted an error? See our corrections policy or email the address on the contact page. Markdown version for agents: /news/inkling-2-small.md
Want these briefs in your Google Top Stories? Add Agentic Weekly as a preferred source on Google.
Newsletter
The Weekly Verdict. Free, every Monday.
A free email every Monday: the week's few big AI releases, then 5 short use cases for the new models, each with a prompt you can paste.
Free. Double opt-in. One click to unsubscribe.