Benchmarks
Long context benchmarks
These benchmarks measure how well models use information in long inputs.
We track 130 benchmarks in this topic. Main benchmarks are listed by how many models they cover, most first. Where a benchmark has separate results for a subset of its tasks, those results appear right under it. Click one for its leaderboard.
Showing 101–130 of 130 benchmarks.
- BABILong qa1 10M3 models
- BABILong qa1 1M3 models
- BABILong qa2 10M3 models
- BABILong qa2 1M3 models
- BABILong qa2 512k3 models
- BABILong qa3 10M3 models
- BABILong qa3 1M3 models
- BABILong qa3 512k3 models
- BABILong qa4 10M3 models
- BABILong qa4 1M3 models
- BABILong qa4 512k3 models
- BABILong qa5 10M3 models
- BABILong qa5 1M3 models
- BABILong qa5 512k3 models
- MMLongBench-128K2 models
- InfiniteBench/En.MC1 model
- InfiniteBench/En.QA1 model
- MRCR 128K (8-needle)1 model
- NIH/Multi-needle1 model
- OpenAI-MRCR: 2 needle 256k1 model