1.1.17 (2026-10-08)
Features
- cli: add --memory-percent and --ram-percent overrides (#1089) (17d45bd)
- tui: add mouse support to existing controls (#1110) (8caf925)
Bug Fixes
- api: make include_too_tight=true return too_tight rows (#1087) (cbd30d7)
- cli: honor --json in search (#1113) (f89395b), closes #901
- cli: reject unknown recommend filter values (#1114) (35fdd4f)
- fit: dense CpuOffload TPS uses measured DDR bandwidth and the GPU/RAM split (#1108) (b3a5974)
- fit: make ranking deterministic across runs (#1066) (f2eafd0)
- fit: pick the best quant within the Marginal ceiling (#1120) (449fe95), closes #1118
- hardware: treat an Intel adapter with no series name as integrated (#1100) (c299bc5)
- models: prefer name-declared size when scraped params are implausible (#925) (f433cf6)
Performance Improvements
- hardware: run lspci once per process (#1065) (0728894)
- tui: optimize model search filtering in apply_filters for 40%+ speedup (#934) (ca864d0)
What's Changed
- docs: new README demo GIF, recorded from a reproducible VHS tape by @AlexsJones in #1068
- perf(hardware): run lspci once per process by @AlexsJones in #1065
- fix(fit): make ranking deterministic across runs by @AlexsJones in #1066
- bench: community results for nvidia-geforce-gtx-1080-ti by @farkusza in #1078
- bench: community results for nvidia-geforce-gtx-1080-ti by @farkusza in #1079
- [Automated] Weekly model data refresh by @github-actions[bot] in #1077
- feat(cli): add --memory-percent and --ram-percent overrides by @MannXo in #1089
- fix(api): make include_too_tight=true return too_tight rows by @SulimanAbdulrazzaq in #1087
- perf(tui): optimize model search filtering in apply_filters for 40%+ by @iam-saiteja in #934
- chore(deps): bump jsonschema from 0.51.0 to 0.55.0 by @dependabot[bot] in #1074
- chore(deps): bump tabled from 0.21.0 to 0.22.0 by @dependabot[bot] in #1073
- fix(models): prefer name-declared size when scraped params are implausible by @DangMinh0412 in #925
- bench: community results for nvidia-geforce-gtx-1080-ti by @farkusza in #1099
- fix(hardware): treat an Intel adapter with no series name as integrated by @Yi-111-a in #1100
- bench: community results for intel-raptor-lake-p-iris-xe-graphics-integrated by @FabioLeitao in #1083
- fix(fit): dense CpuOffload TPS uses measured DDR bandwidth and the GPU/RAM split by @mvanhorn in #1108
- docs: sync Chinese translation with new demo GIF and layout updates by @JasonYeYuhe in #1111
- chore(deps): bump the cargo-minor-patch group across 1 directory with 4 updates by @dependabot[bot] in #1109
- [Automated] Weekly model data refresh by @github-actions[bot] in #1112
- fix(cli): reject unknown recommend filter values by @jayhemnani9910 in #1114
- fix(cli): honor --json in search by @jayhemnani9910 in #1113
- bench: community results for intel-3rd-gen-core-processor-graphics-controller-integrated by @FabioLeitao in #1123
- fix(fit): pick the best quant within the Marginal ceiling by @jayhemnani9910 in #1120
- bench: community results for nvidia-geforce-rtx-4060-laptop-gpu by @celalthedon in #1082
- bench: community results for nvidia-geforce-rtx-4060-laptop-gpu by @jazzwang in #1122
- feat(tui): add mouse support to existing controls by @pschaub in #1110
- bench: community results for amd-radeon-rx-9070-xt by @Siegez in #1088
- bench: reviewed RTX 5070 Laptop GPU measurements by @GerhardBalz-Admin in #1095
- chore(main): release 1.1.17 by @AlexsJones in #1069
New Contributors
- @farkusza made their first contribution in #1078
- @MannXo made their first contribution in #1089
- @iam-saiteja made their first contribution in #934
- @DangMinh0412 made their first contribution in #925
- @Yi-111-a made their first contribution in #1100
- @jayhemnani9910 made their first contribution in #1114
- @celalthedon made their first contribution in #1082
- @jazzwang made their first contribution in #1122
- @pschaub made their first contribution in #1110
- @Siegez made their first contribution in #1088
- @GerhardBalz-Admin made their first contribution in #1095
Full Changelog: v1.1.16...v1.1.17