📊 Full opportunity report: What Is The Real Significance Of Thinking Machines’ Inkling In AI? on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Thinking Machines has publicly released the full weights of its Inkling model under an open license, marking a significant move in AI model openness. However, questions remain about the model’s licensing restrictions and data transparency. This development could influence how AI models are owned, used, and regulated.
Thinking Machines has officially released the full weights of its latest foundation model, Inkling, under an open-source license on Hugging Face. This move is significant because it allows users to download, modify, and deploy the model independently, contrasting with the common practice of proprietary or API-based models. The release underscores a shift toward transparency and ownership in AI development, making Inkling a notable case in the ongoing debate over open versus closed models.
Inkling is a 975-billion-parameter Mixture-of-Experts transformer supporting a 1-million-token context window. It was trained on 45 trillion tokens of text, images, audio, and video, with its multimodal input processed via a novel encoder-free design. The full weights were made available under Apache 2.0 license on Hugging Face, enabling unrestricted download, modification, and ownership of the model. This is a departure from typical industry practice, where models are often licensed with restrictions or kept proprietary.
Despite the open release, reports suggest Thinking Machines maintains a separate Model Acceptable Use Policy (AUP), which may restrict certain applications like surveillance or automated decision-making. Learn more about industry shifts and staffing. The company’s transparency about this layered policy contrasts with the open licensing, raising questions about the true scope of openness and user restrictions. Additionally, the training data and pipeline have not been published, which is a common industry norm but limits full transparency.
The weights came first: what Inkling actually signals
Mira Murati’s lab shipped its first foundation model — and the model isn’t the story. The order of operations is: full weights, Apache 2.0, day one, before any closed API. Plus a rare concession — the lab says it’s not the strongest model available, open or closed.
- AIME 2026 97.1%
- GPQA Diamond 87.2%
- MCP Atlas (Nemotron 44.7%) 74.1%
- VoiceBench · open-weight audio frontier 91.4%
- FORTRESS adversarial · best open 78.0%
- ForecastBench · calibration 61.1
- HLE text-only (GLM-5.2 40.1%) 29.7%
- SWE-bench Pro (GLM-5.2 62.1%) 54.3%
- Terminal-Bench 2.1 (GLM-5.2 82.7%) 63.8%
- SWE-bench Verified (Fable 5 95.0%) 77.6%
- Design Arena · 2nd open, behind GLM-5.2 ~10th
A 0.2 → 0.99 effort setting trades reasoning tokens against cost & latency, so you get a curve, not a point. On Terminal-Bench 2.1 it reportedly matches Nemotron 3 Ultra at ~⅓ the tokens. Peak score is a vanity metric when you serve millions of calls; the cost curve is what ships. (Bonus: its chain of thought compressed on its own during RL — nobody rewarded it; efficiency did.)
Pitched as the Western alternative to Chinese open weights (censorship-resistance training is the differentiator). But GLM-5.2 still wins on agentic/reasoning and Kimi K2.6 often on multimodal: best American open model, second in the open field. The irony — post-training was bootstrapped on synthetic data from Kimi K2.5.
BF16 needs ≥2 TB aggregate VRAM (8× B300 / 16× H200). NVFP4 still needs ≥600 GB. Not a workstation model — a 512 GB fleet falls just short. “Open” ≠ “runnable.” Mitigations: 1-bit GGUFs (~74% acc.), hosted eval routes, and Inkling-Small (12B active) — the release local-first builders actually want.
Open weights used to be a consolation prize. Inkling is a strategic open release — Apache 2.0, natively multimodal, honestly marketed, published complete on day one, optimized for deployment rather than headlines (the model isn’t the product; the fine-tuning platform is). It doesn’t need to win every benchmark for that to matter. The frontier is learning that owning the base beats renting the API — arriving now from the inside. For the sovereignty buyer: ① a real Western hedge against being switched off · ② verify the use policy before you build · ③ check the VRAM, then benchmark vs GLM-5.2 & Kimi K2.6 on your task.
Implications of Open Weights and Usage Policies
This release signifies a major step toward model ownership and transparency in AI, allowing organizations to deploy and adapt Inkling without reliance on API providers. It challenges the industry norm of closed models and could accelerate innovation by enabling independent testing, fine-tuning, and commercialization. However, the potential restrictions embedded in the separate AUP introduce complexities, especially for sensitive domains like public safety or geospatial analysis, where licensing clarity is critical. Overall, this move could influence future industry standards on openness, licensing, and ethical use of large models.
Open Source AI Models On Mobile: Deploying Lightweight LLMs On Android And iOS Devices
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background on Open Model Releases and Industry Norms
Over the past year, several AI labs have moved toward releasing models with open weights, but often with restrictions or limited transparency. Historically, most foundation models have been proprietary, with access limited via APIs. The recent release of Meta’s Llama 2, for example, included open weights but with licensing restrictions. Inkling’s release by Thinking Machines, a startup founded by former OpenAI CTO, marks a notable shift: full weights under a permissive license, yet layered with a possibly restrictive AUP. This approach reflects ongoing industry tension between openness and control, especially amid regulatory and ethical concerns.
Prior to this, most open releases have faced criticism for lack of data transparency and unclear usage boundaries. Inkling’s release, with detailed technical specifications and external benchmark scores, attempts to balance transparency with caution, but the layered AUP complicates the narrative of true open-source access.
“Our goal is to promote transparency while ensuring responsible use through our Acceptable Use Policy.”
— Thinking Machines spokesperson
LOCAL LLM DEPLOYMENT: Training, Fine-Tuning, & Offline Inference: The Complete Developer’s Guide to Building, Training, and Running Private Open-Source AI Offline (with full source code)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unanswered Questions About Inkling’s Licensing and Data Transparency
It remains unclear how enforceable the separate Model Acceptable Use Policy is, and whether it effectively limits certain applications despite the open weights. The specifics of the training data and pipeline have not been disclosed, raising questions about the model’s transparency and reproducibility. Additionally, the impact of these layered restrictions on commercial and research use is still uncertain, especially for organizations in regulated sectors.
LAFVIN ESP32-S3 1.69" LCD Development Board with Camera, AI Vision Voice Development Kit, Programmable IoT Board with Mic Speaker for STEM Education
【Abundant Core Computing Power】 Powered by the ESP32-S3 microcontroller and equipped with a large-capacity memory configuration of 16MB…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps for Industry Adoption and Policy Clarification
Expect further analysis from the AI community on Inkling’s licensing and performance. Organizations interested in deploying the model will need to scrutinize the AUP and assess compliance risks. Industry stakeholders may also push for clearer standards on open model licensing and data transparency, potentially influencing future releases. Monitoring how Thinking Machines and competitors handle licensing and restrictions will be key in understanding the evolving landscape of open AI models.
Laplink PCmover Migration Software – Initial Pay-Per-Use License Fee – Monthly invoicing for additional uses – $29.95/license with Super Speed USB 3.0 cable – Business Technician, 10 Licenses
Flexible Pay-Per-Use Structure: Laplink's Technician licensing bills only for completed transfers. One license covers unlimited transfer attempts from…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What makes Inkling different from other large language models?
Inkling is notable for its full open weights under Apache 2.0, allowing unrestricted download and modification. It also supports multimodal input, including text, images, and audio, with a large context window of one million tokens.
Does releasing the weights mean Inkling is fully open source?
No. While the weights are under Apache 2.0, reports suggest there is a separate Model Acceptable Use Policy that may impose restrictions on use, which complicates the notion of full openness.
Why is the layered licensing and policy important?
It raises questions about enforceability and scope. Users may be restricted from certain applications despite having access to the model weights, especially in sensitive or regulated domains.
What impact could this release have on the AI industry?
This move could accelerate model ownership and transparency, encouraging other labs to release models openly. It may also prompt discussions on licensing standards and responsible AI use.
Source: ThorstenMeyerAI.com
