[AINews] Inkling 2 Small: 17A-594B trained in 68 hours on 12T tokens, beating Muse Spark
- ID
- 43738
- Status
- summarized
- Published
- 06 Nov 2026, 6:35 PM
- Fetched
- 06 Nov 2026, 7:22 PM
- Provider
- Latent Space
- Category
- developer-ai
- Original URL
- https://latent.space/p/ainews-inkling-2-small-17a-594b-trained-in-68-hours-on-12t-tokens-beating-muse-spark/
- Source URL
- https://www.latent.space/feed
Summary
- Score
- 6.5
- Created
- 06 Nov 2026, 7:22 PM
- Tags
- Audience
- developersai_ml_learnersai_agent_users
What happened
Latent Space's AI News brief covers Thinking Machines' Inkling 2 Small, described as a 17B-active/594B-total model pretrained from scratch in under 3 days (68 hours) on under 12T tokens in native NVFP4, followed by a single joint RL run. The post says it is competitive with and often beats same-size Chinese models, that it has a notably strong vision encoder, and that people are calling it the smartest Western open-weight model by a margin. The excerpt also flags a METR study of AI research at Anthropic and quotes John Schulman on the alignment process, but gives no benchmark tables, eval numbers, or pricing.
Why it matters
The only concrete, checkable claims here are the training recipe — 68 hours, <12T tokens, native NVFP4 pretraining, one joint RL run — plus an emphasis on intentional customizability and a strong vision encoder. If you are picking an open-weight base model to fine-tune, that recipe is the thing to probe (can you reproduce a short NVFP4 pretrain plus a single RL pass on your own domain data?), not the 'beats Muse Spark' headline, which the text asserts without any eval numbers. Treat this as a lead to verify against the actual model card and benchmarks before swapping anything in production; the source alone does not justify a migration decision.
Discussion angle
The 68-hour, native-NVFP4, single-joint-RL-run recipe is the interesting part — ask the room whether short, cheap pretraining plus one RL pass changes what a small team could train themselves, and what evidence (benchmarks, license terms, actual FP4 hardware support) they'd need before believing the 'smartest Western open-weight model' claim.