Thinking Machines, the startup co-founded by former OpenAI CTO Mira Murati, has launched Inkling, a multimodal open-weights model with 975 billion parameters. According to Artificial Analysis, Inkling is currently the most powerful U.S. open-weights model, outperforming competitors like Kimi K2.6 and DeepSeek v4 Flash max on agentic tasks while also demonstrating high token efficiency. The model is built as a Mixture-of-Experts Transformer with 975 billion total parameters, 41 billion of which are active at any given time. It is the first model from the startup, which was founded by Murati, who played a key role in developing ChatGPT.
Inkling natively processes text, images, and audio, supporting a context window of up to one million tokens. The weights are freely available on Hugging Face, and the company offers access through Tinker, its platform for adapting AI models to specific tasks. Thinking Machines positions Inkling as a flexible base model for customization, emphasizing its multimodal support, efficient processing, and fine-tuning options. The company pre-trained Inkling on 45 trillion tokens of public and synthetic data, including text, images, audio recordings, and videos. It also used the Chinese AI model Kimi K2.5 to generate synthetic data.
According to AI benchmarking platform Artificial Analysis, Inkling scores 41 on the Artificial Analysis Intelligence Index, making it the leading U.S. open-weights model. It ranks three points above the previous leader, Nemotron 3 Ultra at 38, and well ahead of Gemma 4 31B at 29 and gpt-oss-120b at 24. However, the model performs rather poorly on factual accuracy, scoring +2 on its AA Omniscience benchmark with 40 percent accuracy and a 63 percent hallucination rate. These results are likely to limit its use in applications requiring highly accurate information.
Source: thedecoder
