Understand this story
On November 6, 2026, Thinking Machines Lab released its second-generation model, Inkling 2 Small. According to the official blog, the model uses a MoE architecture with 594B total parameters and 17B activated parameters, supports image input and variable reasoning intensity, and scores 31 on the Artificial Analysis Intelligence Index; it is first offered on Tinker with closed weights, and the full weights are planned to be released under the Apache 2.0 license next week after external safety testing and red-team evaluation are completed. An Artificial Analysis article states that its score is 18 points higher than Inkling Small, on par with Gemini 3.8 Flash (high), and that it is priced at $0.30/$1.00 per 1M input/output tokens. On the same day, Arena announced that Inkling 2 Small has entered Code Arena: WebDev with an AutoEval score of 1654, and stated that once its weights are released it will rank 1st among US open models, 4th among open models worldwide, and 20th on the overall leaderboard, an improvement of 245 points over the previous generation Inkling Small's 1409.
AI-generated from public reports
Latest development · Report
Thinking Machines releases Inkling 2 Small, scoring 31 on the AA Intelligence IndexDevelopments
2 confirmed developments
- Source:English text is being prepared.
- Source:Arena
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 6
Nathan Lambert
Thinking Machines releases Inkling 2 Small, scoring 31 on the AA Intelligence Index
Thinking Machines has released Inkling 2 Small, which scores 31 on the AA Intelligence Index, a notable improvement over its predecessor. The model is now available for fine-tuning and inference on Tinker, with full weights planned for release soon. Nathan Lambert commented that this may be the smartest Western open-weight model currently available; it is smaller in scale than muse spark open and mistal large, but uses the more permissive Apache license.
Soumith ChintalaRelated discussion
Soumith Chintala joined this discussion
This post contributes to heat; it is not a formal report or confirmed fact.
TeortaxesRelated discussion
Teortaxes joined this discussion
This post contributes to heat; it is not a formal report or confirmed fact.
Artificial AnalysisSelected
Thinking Machines releases Inkling 2 Small, scoring 31 on the Intelligence Index
Thinking Machines releases the reasoning model Inkling 2 Small, which scores 31 on the Artificial Analysis Intelligence Index v5.0.1, 18 points higher than Inkling Small and on par with Gemini 3.8 Flash (high). The model is initially offered as closed-source through the Tinker platform and Tinker Playground, with plans to later open its weights under the Apache 2.0 license; it has 594B total parameters and 17B activated parameters, priced at $0.30/$1.00 per 1M input/output tokens. Evaluations show notable gains on agentic tasks, with AutomationBench-AA rising from 5% to 46% and FrontierCode 2.0 from 5% to 25%, but Terminal-Bench 4.0 stands at 13%, below Claude Sonnet 5.5 (medium)'s 30%; average output per task is about 150K tokens, at a cost of $1.18, roughly 6 times that of Inkling Small.
Arena
Thinking Machines' Inkling 2 Small enters Code Arena WebDev…
Arena announced that Thinking Machines' Inkling 2 Small has entered Code Arena: WebDev with an AutoEval score of 1654. Arena stated that once its weights are released, it will rank 1st among US open-source models, 128 points higher than 2nd-place Meta Muse Spark Open, 4th among global open-source models and 20th on the overall leaderboard, a 245-point improvement over its predecessor Inkling Small's 1409. The model is priced at $1.00/1M for output and $0.30/1M for input. Thinking Machines said fine-tuning and inference are already available on Tinker, with full weights planned for release soon.
Mira Murati
Thinking Machines releases small model Inkling 2 Small, AA Intelligence…
Thinking Machines releases the small model Inkling 2 Small, which scores 31 on the AA Intelligence Index, a significant improvement over the previous generation. The model is now available on Tinker for fine-tuning and inference, with full weights planned for release soon. Mira Murati reposted the news.
Thinking Machines Lab: Official BlogFirst-partySelected
Thinking Machines Lab Releases Inkling 2 Small, a 594B Total Parameter MoE Model
Thinking Machines Lab has released its second-generation model Inkling 2 Small, which uses a 594B total parameter, 17B active parameter MoE architecture, supports image input and variable reasoning effort, and scores 31 on the Artificial Analysis Intelligence Index. The model is first offered on Tinker with closed weights, and is planned to be released with full weights under the Apache 2.0 license next week after external safety testing and red-team evaluation are completed. The company says it delivers comprehensive improvements over the previous generation in long-horizon agentic coding, vision, and knowledge work, and its training includes 8.4 trillion token pretraining and joint RL with 1075 steps and over 17 million rollouts.
English text is being prepared.Related discussion
English text is being prepared. joined this discussion
This post contributes to heat; it is not a formal report or confirmed fact.
Nov 5
Full Artificial Analysis articleFirst-party
Thinking Machines releases Inkling 2 Small, scoring 31 on the Intelligence Index
Thinking Machines releases Inkling 2 Small, which scores 31 on the Artificial Analysis Intelligence Index v5.0.1, 18 points higher than Inkling Small and on par with Gemini 3.8 Flash (high). The model is currently proprietary, offered through the Tinker platform and Tinker Playground, with plans to later open its weights under the Apache 2.0 license. The model has 594B total parameters and 17B activated parameters, supports text and image input, and is priced at $0.30/$1.00 per 1M input/output tokens.