Thinking Machines Lab, the startup founded by former OpenAI CTO Mira Murati, released Inkling 2 Small on Nov. 5, the first model in its second generation [1]. It is a mixture-of-experts model with 594 billion total parameters, 17 billion of them active per token, plus a 292-million-parameter vision encoder [1]. The model is available now for sampling and fine-tuning on Tinker, the company’s training platform, and Thinking Machines says it will release the full weights under an Apache 2.0 license next week, after external safety testing and red-teaming [1].
The launch landed as Business Insider reported that the company is raising $5 billion in a round led by Accel at an expected $40 billion valuation, according to SiliconANGLE [2]. Thinking Machines has not commented publicly on the round. In September, The Information reported talks at $5 billion to $6 billion on a pre-money valuation of at least $40 billion, with Nvidia expected to supply about half [3]. The lab raised $2 billion at a $12 billion post-money valuation in July 2025 [3].
Built for coding and “autoresearch”
Thinking Machines pretrained the model on 8.4 trillion tokens in 68 hours on Nvidia GB300 NVL72 racks, then ran a single nine-day reinforcement learning run on 2,048 GPUs across coding, agent and reasoning tasks [1]. The company pitches it at “autoresearch” work such as post-training other models and optimizing GPU kernels. In one test it fine-tuned Qwen3.5-4B for time-series forecasting on Tinker in about 4.5 hours, spending $426 of a $500 budget and lifting the model’s skill score from 0.191 to 0.289 [1].
On the company’s own benchmark table, Inkling 2 Small scored 72.1% on DeepSWE v1.1, behind DeepSeek-V4.1-Flash at 77.0% and ahead of Kimi K3 at 68.5%, and led the listed open models on MMMU Pro (81.6%) and ScreenSpot Pro (83.8%) [1]. The company says it still trails closed-weights models on alignment and that its chat experience is weaker, including overreacting to mental-health and safety queries [1].
MLQ Newsletter
At the intersection of AI, tech, and markets.
The stories that matter, in one email. Free — unsubscribe anytime.
You’re subscribed!