EN · International edition · 16/10/2026
Inferama daily · 16/10/2026
The most important AI developments since the previous newsletter, selected for editorial relevance.
News coverage period: 15/10/2026 16:05 → 16/10/2026 08:30 (Europe/Madrid).
Documentary audit of Ai2’s open-model catalogue: what the licensing evidence supports
An assessment of the primary documentation available to verify the code, weight, and data licenses for Ai2 models, highlighting current gaps in legal traceability.
Read article →Did the Agent Really Finish? How to Verify the Result Without Trusting Its Final Answer
An action being issued, a tool responding, and a task being completed are different things. Learn to define an observable result, choose appropriate evidence, and communicate precisely what has been verified.
Read article →Nemotron 3.5 Super VL: NVIDIA Announces a 120B-Parameter Multimodal Model
The new model combines reasoning, tool use, and understanding of text, images, documents, and video. NVIDIA says it activates 12 billion parameters per token and offers efficiency improvements, but the published results are not enough to establish an independent comparison or a universal inference cost.
Read article →How to Verify AI Claims: A Protocol for Deciding What to Keep, Correct, or Remove
A citation does not prove that a claim is accurate. Learn to separate verifiable claims, read evidence in context, and record a decision another person can reproduce.
Read article →Before Automating with AI: How to Evaluate a Task and Approve a Pilot
A practical protocol for defining a task, comparing it with the current process and a deterministic alternative, and setting acceptance, human-review, and stop criteria before testing.
Read article →Persistent Memory for Agents: What to Store, for How Long, and How to Correct It
A practical design guide to memory records, lifecycle policies, and verification tests for AI agents—distinguishing persistent memory from conversation history and temporary task state.
Read article →Aleph Alpha and Cohere: What the Merger Means for Access to Sovereign Models—and What Still Needs Verification
An analysis of the Aleph Alpha–Cohere combination: confirmed facts, potential effects on sovereign model availability, regulatory implications, and a checklist for verifying licences and access channels.
Read article →What a study on looped Transformers can—and cannot—show about OpenAI
An experimental paper describes a way to stabilize looped Transformers and measure their performance. It does not show that OpenAI uses this approach in its next model, nor does it establish that the architecture makes safety oversight harder.
Read article →Muse Spark Open and Mac Studio: What Is Confirmed About Meta’s Model
Muse Spark Open is attributed with 816 billion parameters, 42 billion active parameters, and a one-million-token context window. However, the available primary documentation from Meta does not confirm this variant or official Apple Silicon optimization; local hardware requirements and performance also remain unverified.
Read article →NVIDIA unveils Nemotron 3.5 Super VL, a 120-billion parameter open model for enterprise agents
Nemotron 3.5 Super VL is designed to interpret documents, charts, images, and video while supporting long-running tasks with tools. Available information confirms the launch and model scale, but it is insufficient to establish independent performance or the specific NVIDIA hardware required for each deployment.
Read article →