Inkling: Inside Thinking Machines Lab's First Open-Weight AI Model

Release platform and accessibility: Inkling's full weights were made available on July 15, 2026. The model is available for fine-tuning on Thinking Machines' own Tinker platform, and the company states users can "play with it" on an interactive interface called the Inkling Playground . Because the weights are open, external deployment is also possible.

Modalities: Inkling reasons across text, image, and audio modalities .

Strategic positioning against closed-model providers: Thinking Machines Lab explicitly positioned Inkling as "a significant departure from the flagship offerings of OpenAI, Anthropic, or Google" by releasing it as an open-weight model . The company is doubling down on transparency, customizability, and developer freedom — contrasting with the closed APIs and gated access of the major frontier labs. This is consistent with the company's stated mission of making advanced AI systems "more widely understood, customizable, and generally available" .

Company Funding & Public Benefit Structure

Funding: Thinking Machines Lab raised $2 billion in seed funding — the largest seed round in venture capital history — led by Andreessen Horowitz, with participation from Nvidia, AMD, Cisco, Accel, Jane Street, and others . Valuation was reported between $10 billion and $12 billion at close .

Public benefit corporation: The company is structured as a Public Benefit Corporation (PBC), which legally requires it to balance shareholder value with a stated public benefit purpose . Its mission emphasizes reliability, transparency, interpretability, and human-AI collaboration .

Competitive Landscape

Inkling enters a market crowded with strong open-weight models — including Meta's Llama 4, Mistral's models, and DeepSeek's open releases — as well as closed frontier models from OpenAI (GPT-5), Google (Gemini 2.0), and Anthropic (Claude 4). Key differentiators for Thinking Machines Lab include:

• Founding team pedigree: Led by former OpenAI CTO Mira Murati, with ex-OpenAI research talent including co-creators of RLHF and the AI researcher who built the safety culture that shaped GPT-4 .
• Tinker ecosystem: Inkling is natively integrated with Tinker, a low-level fine-tuning platform that abstracts distributed GPU orchestration while giving developers full control over training loops — a narrower, more developer-focused pitch than the broad consumer-facing products of OpenAI or Google .
• Open-weight + PBC framing: The combination of full-weight openness with a legal public benefit structure is rare at this scale, potentially appealing to enterprise customers and regulators concerned about concentrated AI power .
• Early stage caveat: Inkling is the company's first proprietary model release. Independent benchmarks comparing its raw capabilities against GPT-5, Claude 4, or Llama 4 were not yet available at launch .