Artificial Analysis launches Optima, a tool for benchmarking AI models on your own data
New custom benchmarking platform compares not just quality but cost and time per task
Current rank (1M)
13
Rank over the last 25 days
2026.09.14 – 2026.10.08 · High 13 · Low 52 · Now 13
New custom benchmarking platform compares not just quality but cost and time per task
WSJ reports Apple proposed a usage-based payment model, paying each time content is used, with a budget in the hundreds of millions of dollars
At the Ai4 conference, three giants of the field offered differing takes on the risks of open-weight models and concerns over gatekeeping
Apple unveils MoMo, a robot manipulation framework that controls whether the same motion is performed quickly or cautiously
Apple analyzes UMAP's neighbor graph directly, bypassing the usual visualization, to surface representative points and dense structures
Apple Research presents a method to create preference data without extra annotation or external models
"DLR-Lock" research lets users run open-weight models as-is while blocking modification
Analysis finds diffusion language models lag in long-context tasks while autoregressive models pull ahead in batch processing
Trained on 2.1 trillion tokens, then self-distilled to generate sentences in as few as 4 steps, testing an alternative to autoregression
That's the last story.