Announcing Artificial Analysis Openness Index

'Open weights' hides differences in license, training data and methodology disclosure. A standardized rubric helps engineers judge what they can host, fine-tune or replicate.

• pretraining — The first, biggest phase of building a model: training it on enormous amounts of text so it learns language, facts, and reasoning in general.
• open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
• fine-tuning — Taking a trained model and training it a bit more on your own examples so it gets better at one specific job.

• articleAnalysis Openai Gpt Oss Models
• articleGpt 5 Benchmarks And Analysis
• articleKimi K2 Thinking Everything You Need To Know
• articleAa Omniscience Knowledge Hallucination Benchmark

• articleThe Open-Model Race has Split Four WaysAlphaSignalAI
• articleOpen-Source AI & Open Models Reading ListNathan Lambert
• articleThe current balance of power in open modelsNathan Lambert

Checking sign-in…

Loading comments…