1.1.13 (2026-09-03)
Features
- add --llama-cpp-path global CLI flag mirroring LLAMA_CPP_PATH env var (#583) (3222da5)
- docker: add multi-stage build (#911) (ed8c184)
- docker: add web frontend service and multi-stage container support (ed8c184)
- hardware profiles, MoE Tier-2 fixes, and estimate confidence (#969) (#971) (a8a1a93)
Bug Fixes
- cli: return JSON errors for missing models (#966) (1a147ed)
- hardware: recover BIOS UMA carveout on Linux APUs and drop legacy Intel iGPUs (#964) (#995) (bb01690)
- models: keep architecture metadata when config.json fetch misses (#963) (ec652ba)
- plan: honor hardware profile bandwidth (f9d8e48)
- providers: find LM Studio models on disk, not just loaded ones (#902) (413246f)
- providers: identify vLLM, LM Studio and Docker Model Runner by evidence (#984) (b624527), closes #791 #790
- providers: unify the two MLX quant-suffix strippers (#895) (02c2fa5), closes #869
- providers: verify server identity before importing models (#791) (#938) (9319453), closes #790
- quality: disable Ollama thinking for benchmarks (#975) (eb11c9e)
- quality: preserve benchmark regex escapes (#967) (4f8b489)
- share: explain a refused write to the user's own fork (#865) (d6bfa10), closes #862
What's Changed
- bench: community results for nvidia-geforce-rtx-5090-laptop-gpu by @sleggett-ai in #913
- bench: community results for amd-radeon-rx-6800 by @nantobreak in #915
- bench: community results for apple-m3-max by @aikrash in #954
- bench: community results for nvidia-geforce-rtx-4060-laptop-gpu by @jabbadabbadahat in #958
- chore(deps): bump astral-sh/setup-uv from 9.0.0 to 10.0.1 by @dependabot[bot] in #944
- chore(deps): bump jsonschema from 0.47.0 to 0.49.4 by @dependabot[bot] in #905
- chore(deps): bump base64 from 0.22.1 to 0.23.1 by @dependabot[bot] in #906
- test(hardware): add nvidia-smi parser fixtures by @luojiyin1987 in #940
- test(hardware): add system_profiler parser fixtures by @luojiyin1987 in #941
- test(hardware): add rocm-smi parser fixtures by @luojiyin1987 in #942
- fix(providers): unify the two MLX quant-suffix strippers by @Akciali in #895
- fix(share): explain a refused write to the user's own fork by @AlexsJones in #865
- [Automated] Weekly model data refresh by @github-actions[bot] in #945
- fix(models): keep architecture metadata when config.json fetch misses by @AlexsJones in #963
- feat: hardware profiles, MoE Tier-2 fixes, and estimate confidence (#969) by @saman-mb in #971
- fix(quality): preserve benchmark regex escapes by @mikemikimike in #967
- bench: community results for radeon-rx-9060-xt by @s7a5a1 in #965
- bench: community results for amd-gpu by @Sky550 in #977
- bench: community results for apple-m4 by @Takahiro3D in #976
- fix(quality): disable Ollama thinking for benchmarks by @mikemikimike in #975
- fix(cli): return structured JSON errors for missing models by @mikemikimike in #966
- docs: sync Chinese README with 1.0 restructured documentation by @JasonYeYuhe in #982
- bench: community results for nvidia-geforce-rtx-4070-ti by @fatinghenji in #983
- [Automated] Weekly benchmark cache refresh by @github-actions[bot] in #981
- chore(deps): bump docker/setup-buildx-action from 4.2.0 to 4.3.0 in the github-actions-minor-patch group by @dependabot[bot] in #980
- chore(deps): bump rmcp from 2.2.0 to 3.1.0 by @dependabot[bot] in #979
- chore(deps): bump the cargo-minor-patch group with 2 updates by @dependabot[bot] in #978
- fix(providers): verify server identity before importing models by @Akciali in #938
- feat(docker): add multi-stage container build by @lj020326 in #911
- fix(providers): find LM Studio models on disk, not just loaded ones by @ba1lly in #902
- feat: add --llama-cpp-path global CLI flag mirroring LLAMA_CPP_PATH env var by @mvanhorn in #583
- bench: community results for intel-r-arc-tm-b390-gpu by @ynaoHYDE-dev in #986
- bench: community results for intel-coffeelake-s-gt2-uhd-graphics-630-integrated by @lopez848 in #987
- fix(providers): identify vLLM, LM Studio and Docker Model Runner by evidence by @Akciali in #984
- Fix Score sort direction floating unrunnable models to the top by @psylsph in #991
- Add llmfit-gui project to README by @raiyyan729-cloud in #990
- bench: community results for amd-ati by @haylocki in #988
- bench: community results for nvidia-geforce-gtx-1650-with-max-q-design by @koloroner in #985
- fix(hardware): recover BIOS UMA carveout on Linux APUs and drop legacy Intel iGPUs by @AlexsJones in #995
- chore(main): release 1.1.13 by @AlexsJones in #962
New Contributors
- @sleggett-ai made their first contribution in #913
- @nantobreak made their first contribution in #915
- @aikrash made their first contribution in #954
- @jabbadabbadahat made their first contribution in #958
- @saman-mb made their first contribution in #971
- @mikemikimike made their first contribution in #967
- @s7a5a1 made their first contribution in #965
- @Sky550 made their first contribution in #977
- @Takahiro3D made their first contribution in #976
- @fatinghenji made their first contribution in #983
- @lj020326 made their first contribution in #911
- @ynaoHYDE-dev made their first contribution in #986
- @lopez848 made their first contribution in #987
- @psylsph made their first contribution in #991
- @raiyyan729-cloud made their first contribution in #990
- @haylocki made their first contribution in #988
- @koloroner made their first contribution in #985
Full Changelog: v1.1.12...v1.1.13