AA Long Context Reasoning
A bird's-eye view of the benchmark: what it measures and every AI IQ chart built on it.
About AA Long Context Reasoning
What AA Long Context Reasoning measures
Archived benchmark. Superseded by AA-LCR v1.1 with corrected grading. These original scores are excluded from IQ and are not interchangeable with v1.1.
Reading the cost view
AA Long Context Reasoning publishes no per-run cost, so this chart falls back to each model's estimated effective cost per 1M I/O tokens (a runtime estimate from measured or imputed token usage, not a published task cost). Explicit reasoning/nonreasoning siblings are connected where both are public; every other model is a standalone point. Up and to the left is better.
Reading the completion-time view
Each point is a public model's AA Long Context Reasoning score against its estimated completion time (time to last token) for the input and output lengths set by the sliders. Timing comes from the site's per-model speed measurements, not from the benchmark run, so every point is an estimate and labels carry a ~ prefix. The newest generation with a published result is shown by default. Up and to the left is better.
Reading the score bars
Bars rank models by the source-backed benchmark value used for this chart. Longer bars indicate higher published scores.