news.volyx.in

Inkling: Our Open-Weights Model (thinkingmachines.ai)

1227 points by vimarsh6739 · 43 days ago · 294 comments on HN

Article summary

Thinking Machines has released a new open-weights model called Inkling, which is a Mixture-of-Experts transformer with 975B total parameters and supports a context window of up to 1M tokens. Inkling is designed to be a broad, balanced foundation model that can be fine-tuned for specific tasks and has been trained on a large dataset of text, images, audio, and video. The model is available for fine-tuning on the Tinker platform, and its multimodal capabilities allow it to reason over text, images, and audio. Inkling is the first in a family of models of different sizes, with a smaller version called Inkling-Small also available.

Main themes

  • Open-weights models
  • AI model development
  • Fine-tuning and customization
  • Multimodal capabilities
  • Business models for AI companies
  • Global competition in AI

What commenters say

  • The release of Inkling is a significant development in the field of open-weights models, and its performance and capabilities are being compared to other models such as GLM 5.2.
  • Some commentators believe that the US needs its own open-weights model to compete with Chinese companies like DeepSeek and Z.ai, while others argue that this is not a necessary or desirable goal.
  • The business model of giving away open-weights models and charging for fine-tuning and other services is seen as a viable approach by some, but others are skeptical about its long-term sustainability.
  • There are concerns about the potential for government interference in the development and release of open-weights models, particularly in countries with more authoritarian regimes.
  • The importance of multimodal capabilities in AI models is highlighted, with some commentators arguing that this is a key advantage of Inkling over other models.
  • The benchmarks used to evaluate AI models are seen as limited and not always reflective of real-world performance, with some commentators arguing that other factors such as flexibility and determinism are also important.