📊 Full opportunity report: From Inkling To Innovation: What AI’s Early Signs Mean on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Thinking Machines released its first foundation model, Inkling, under an open license, emphasizing transparency but not claiming to be the strongest. The move highlights ongoing debates about open AI models and ownership costs.
Thinking Machines has officially released its first foundation model, Inkling, under an open license on Hugging Face. This marks a significant development in AI transparency, as the model’s weights are available for download and modification, but it is not claimed to be the most powerful model currently available.
Inkling is a 975-billion-parameter Mixture-of-Experts transformer supporting a 1-million-token context window. It was pretrained on 45 trillion tokens of diverse data, including text, images, audio, and video, with a native multimodal input design. The model is released under Apache 2.0 license, allowing users to download, modify, and deploy it independently.
Despite the open weights, the training data and pipeline are not publicly disclosed, and reports suggest that Thinking Machines maintains a separate acceptable use policy restricting surveillance, deception, and automated decision-making affecting individuals. This introduces a layer of restrictions layered over the open licensing, which has prompted discussions about true openness and control.
Independent benchmark scores from external sources indicate Inkling performs well in areas like speech and safety benchmarks but is mid-tier or behind in some language understanding tests, such as Humanity’s Last Exam. The model’s safety features are notably strong compared to some closed models, especially in refusal tasks.
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-Source AI Ownership
The release of Inkling under an open license represents a notable shift toward greater transparency and user control over large AI models. It allows organizations and developers to fine-tune, inspect, and deploy the model independently, reducing reliance on proprietary APIs. However, the layered restrictions via the acceptable use policy raise questions about the true openness of the model, especially for sensitive applications like surveillance or public safety. This development could influence future licensing practices and the broader debate over AI transparency and regulation.
The MCP Standard: A Developer's Guide to Building Universal AI Tools with the Model Context Protocol
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background on Open Model Releases and Industry Norms
Until now, most large foundation models have been released with restricted access, either through closed APIs or limited open weights, often with licensing that restricts commercial use or modification. The recent release of models like Meta’s Llama 2 and OpenAI’s GPT variants reflects a growing trend toward open access, but with varying degrees of transparency. Inkling’s release under Apache 2.0, combined with the layered use restrictions, marks a nuanced approach that balances openness with control, reflecting ongoing industry debates about ownership, safety, and regulation.
“Our goal is to foster transparency while maintaining responsible use. The layered policy helps us ensure the model is used ethically.”
— Thinking Machines spokesperson
multimodal AI model with large context window
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unclear Aspects of Licensing and Use Restrictions
It remains unclear how strictly the acceptable use policy will be enforced and whether it effectively limits certain applications despite the open weights. The extent to which users can freely modify or commercialize the model without violating the policy is still under scrutiny, and legal interpretations may vary across jurisdictions.
AI in Content Moderation: Automating Online Safety with Artificial Intelligence: Strategies and Tools for Ethical and Effective AI-Powered Online … (Tech Horizons: Your Gateway to Innovation)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps in Model Adoption and Industry Impact
Expect further independent testing and benchmarking of Inkling’s performance and safety features. Industry observers will monitor how organizations adopt the model, whether the layered restrictions influence usage, and if other developers follow suit with open licenses combined with use policies. Regulatory discussions around transparency and control are likely to intensify as more open models emerge.
NVD RTX PRO 6000 Blackwell Professional Workstation Edition Graphics Card for AI, Design, Simulation, Engineering – 96GB DDR7 ECC Memory – 4th Gen RT/5th Gen Tensor Core GPU – OEM Packaging
[NVIDIA Blackwell Streaming Multiprocessor] The new SM features increased processing throughput, and new neural shaders that integrate neural…
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 a 975-billion-parameter multimodal model released under an open Apache 2.0 license, allowing broad access and modification, but with layered restrictions via an acceptable use policy.
Does open weights mean the model is fully open source?
No. While the weights are open under Apache 2.0, the training data, pipeline, and usage restrictions are not publicly disclosed, which complicates the open source classification.
Why does the layered use policy matter?
The policy may restrict applications like surveillance, deception, or automated decision-making, which could limit how freely the model can be used despite the open weights.
What are the risks of using an open model with layered restrictions?
Legal and ethical uncertainties may arise, especially regarding compliance with the use policy and the potential for misuse if restrictions are not clearly enforced.
Source: ThorstenMeyerAI.com
