Inkling (Thinking Machines)
Open-weights multimodal model, ready to tinker
About Inkling (Thinking Machines)
Inkling is an open-weights multimodal model from Thinking Machines Lab that accepts text, image and audio inputs. Built on a Mixture-of-Experts architecture (975B total / 41B active parameters) with a 1M-token context window, it is a generalist strong at knowledge, math, science and coding. Released with open weights, it is designed to be fine-tuned and adapted for domain-specific applications, agentic coding and multimodal — including audio — processing.
Features
- Open weights for self-hosting and fine-tuning
- Multimodal input: text, image and audio
- Mixture-of-Experts (975B total / 41B active parameters)
- 1M-token context window
- Strong knowledge, math, science and coding performance
- Suited to agentic coding and tool use
- Calibrated-confidence forecasting
Who Is It For?
- AI researchers and ML engineers
- Teams fine-tuning models for domain-specific tasks
- Builders of agentic and multimodal applications
Use Cases
- Fine-tuning for domain-specific applications
- Agentic coding and tool use
- Multimodal processing across text, image and audio
- Self-hosting an open-weights model
FAQ
Is Inkling open source? It's released with open weights for self-hosting and fine-tuning.
What inputs does it support? Text, image and audio.
How big is it? 975B total parameters with 41B active (Mixture-of-Experts) and a 1M-token context window.
Build and fine-tune on open weights with Inkling.