TL;DR — Key Takeaways
- Thinking Machines released Inkling 2 Small, a second-generation AI model with 594 billion total parameters, available through its Tinker platform.
- Artificial Analysis found major improvements in agentic benchmarks, although Inkling 2 Small costs nearly six times as much per benchmark task as its predecessor.
- Thinking Machines is reportedly nearing a $5 billion funding round that would value the company at approximately $45 billion after the investment.
Thinking Machines Lab has released Inkling 2 Small, a second-generation AI model designed for autonomous coding, visual reasoning and knowledge work. The company, founded by former OpenAI CTO Mira Murati, is also reportedly nearing a $5 billion funding round.
Introduced Thursday, Inkling 2 Small uses a mixture-of-experts architecture with 594 billion total parameters and 17 billion active parameters. The model supports image inputs and variable reasoning effort. According to Thinking Machines, it can also use visual feedback to build and refine websites, games and 3D designs.
Inkling 2 Small is available through the company’s Tinker platform for inference and fine-tuning, as well as its online Playground. The model initially has closed weights, but Thinking Machines said it plans to release them under an Apache 2.0 license next week, following external safety testing and red-teaming.
Thinking Machines said it made several changes to how it trained Inkling 2 Small. The initial pretraining phase processed 8.4 trillion tokens of text, code and image data in 68 hours using Nvidia GB300 NVL72 systems, followed by another 3.8 trillion tokens of newer data. For reinforcement learning, the company combined reasoning and agentic tasks into a single training run, instead of training them separately as it did with the previous Inkling model. That nine-day run used 2,048 GPUs and completed more than 17 million rollouts across tasks including software engineering, document creation, visual analysis and tool use.
Independent testing found substantial performance gains over the previous Inkling model. Artificial Analysis gave Inkling 2 Small a score of 31 on its Intelligence Index, an 18-point improvement over its predecessor, with the largest gains in agentic tasks. Its score on AutomationBench-AA, which tests automated business workflows, rose from 5% to 46%, while its FrontierCode 2.0 score for software engineering increased from 5% to 25%.
Results were more mixed against competing models. Inkling 2 Small outperformed Claude Sonnet 5.5 (medium) on FrontierCode 2.0 but trailed it on Terminal-Bench 4.0. The improved performance also came at a higher cost. Artificial Analysis found that Inkling 2 Small generated roughly four times as many output tokens per Intelligence Index task, bringing the average cost to $1.18, compared with $0.20 for Inkling Small.
Separately, Business Insider reported that Thinking Machines is nearing a $5 billion funding round led by venture capital firm Accel at a $40 billion pre-money valuation. Citing multiple sources familiar with the negotiations, the publication said the financing could close as soon as this week. If completed on those terms, the investment would give Thinking Machines a post-money valuation of approximately $45 billion, nearly four times the $12 billion valuation attached to its $2 billion seed round in July 2025. Thinking Machines declined to comment, while Accel did not respond to a request for comment.
That initial investment came nearly three months before Thinking Machines launched its first product, the Tinker model fine-tuning platform. Investors were largely betting on Murati and a team of experienced AI researchers, many recruited from OpenAI. The company has since begun developing its own AI models while expanding its compute. In March, Thinking Machines announced a partnership with Nvidia to deploy at least one gigawatt of Vera Rubin systems beginning in 2027, along with an undisclosed Nvidia investment. Inkling 2 Small gives Thinking Machines another technical achievement to show for its substantial investment, although the company has disclosed little about its revenue or financial performance thus far.
