Version 92ff3f3: release notes, benchmarks measured again on November 15
Agent session measured on 14d173b; llama.cpp at a smaller batch on the deep 27B prompts
Benchmarks: current llama.cpp 1.0.0 results only
Drop the retired small-model comparison
Version 14d173b: release note, Gemma 4 results, current pages
Benchmarks: 253,000-token row, 9B table at a 524,288-token context
Version afaf544: release notes, benchmarks against llama.cpp 1.0.0, download
Version 5db82d6: release notes, benchmarks of 11/13/2026, download from the site
One current version, 1123d04, on every page that names one
Pages of 2026-11-10
LexiSuite wiki, 2026-11-07 (source e1ffc87e)