Headline
⚠️ These notes are AI generated and will be revised by a human ⚠️
- New container backends for AMD Strix Halo — DS4, FP4, Halogen Flash, Gufo, and llama.cpp — let you run more models on the platform.
- TheNoise image backend now runs on NVIDIA GPUs for Linux x86_64 alongside ROCm.
- SystemOne decision models and GGUF-based zero-shot classifiers add new routing and decision-making capabilities.
lemonade bench can submit results to the online leaderboard at bench.lemonade-server.ai.
Breaking Changes
⚠️ These notes are AI generated and will be revised by a human ⚠️
- The
/models API response no longer includes image_defaults or audio_defaults keys; their values are now under recipe_options — clients reading those fields must update their read path.
- Router residency conflicts now return HTTP 409 (
router_residency_conflict) instead of HTTP 400 — error handlers expecting 400 will need updating.
- Router policy labels are now strictly validated (unique, non-blank, no control characters); a saved policy with invalid labels will fail to load at startup and all requests return 500 — re-save or edit the policy to use valid labels.
- The
enable_containers config key was removed; any user or script that reads or sets it will encounter an unknown config key — remove it from your configuration.
- FLM chat behavior changed:
max_tokens now returns the exact requested count (one fewer than before), streamed usage is opt-in via stream_options.include_usage, and invalid requests that were silently ignored now get HTTP 400 — adjust clients that depend on the old lenient behavior.
- The
/files endpoint for a multi-shard GGUF model in extra_models_dir now returns one entry per shard instead of one summary entry — clients iterating the response and expecting a single main-file entry may need updating.
Lemonade Server
Other platforms? See our Installation Options for Docker, Snap, Arch, Debian, and more.
Embeddable Lemonade
Portable binaries for bundling into your own installer. Run lemond ./ as a subprocess.
What's Changed
Thanks @AaronStGeorge, @Bekhouche, @DigitalFlow, @MrAnayDongre, @Zorro-Zhang, @bitgamma, @danielholanda, @festeraeb, @fl0rianr, @github-actions, @iswaryaalex, @jeremyfowers, @julianxhokaxhiu, @kenvandine, @meghsat, @mike2307, @pwilkin, @sawansri, @sreeram-11, @superm1 for your awesome contributions to this release!
Click to expand changelog
- add --submit/--submit-file to bench by @bitgamma in #3284
- fix(gui): missing translation gaps by @fl0rianr in #3990
- fix(app): use server sync for individual and bulk model updates by @fl0rianr in #3957
- grabbing rocm versions per build by @iswaryaalex in #3996
- fix(gui): show chat readiness and automatic backend install progress by @fl0rianr in #3992
- Adding zero-shot classifier support by @meghsat in #3952
- feat(backends): add TheNoise CUDA support by @fl0rianr in #4001
- fix(app): support Firefox chatbot sidebar prompts by @fl0rianr in #3973
- fix(gui3): Chat showing the wrong model after loading one from the Models tab (issue #4011) by @meghsat in #4013
- fix(server): add missing backend args by @fl0rianr in #4021
- fix(server): Turning off saved chats will retain the already saved chat sessions by @meghsat in #4012
- fix(gui3): honor saved ctx_size when loading custom models (#4019) by @mike2307 in #4020
- fix(server): apply default_model_source to built-in model downloads by @Zorro-Zhang in #4002
- fix(gui3): Fixed bugs in Model load banners by @meghsat in #4030
- refactor(models): list every model file in server_models.json by @jeremyfowers in #4041
- chore: bump pwilkin backend versions by @pwilkin in #3993
- feat(images): support optional upscale factor for TheNoise by @fl0rianr in #4046
- feat(backends): run container backends through ContainerManager, starting with llamacpp:nathanw by @jeremyfowers in #3709
- feat(ds4): move ds4:rocm to a container backend by @jeremyfowers in #3710
- feat(rocmfpx): add the ROCm FPX container backend by @jeremyfowers in #3711
- feat(halogen): add the Halogen Flash container backend by @jeremyfowers in #3712
- ci: keep container backend images on the newest build of their tags by @jeremyfowers in #3713
- ci: Add cuda backend validation for llama.cpp on Linux using self-hosted amd64 and arm64 runners by @kenvandine in #3714
- fix(gui): align chat toolbar controls (#4051) by @fl0rianr in #4053
- feat(llamacpp): add the halobox container backend by @jeremyfowers in #4029
- feat(gufo): add the Gufo container backend by @jeremyfowers in #4031
- Add SystemOne decision models: /v1/systemone, GGUF zero-shot on /v1/classify, and the systemone router classifier by @meghsat in #3956
- Remove
image_defaults and audio_defaults by @bitgamma in #4049
- Add gfx117x ROCm targets by @superm1 in #4057
- Fixes the GUI3 banner in the news page by @meghsat in #3959
- docs(api): re-record API examples for llama.cpp b12109 by @jeremyfowers in #4064
- Add GUI overhaul video to news page by @jeremyfowers in #4058
- feat(backend): FLM 1.0.11 decision models support by @meghsat in #4037
- test(mcp): cover model aliases in gateway tool calls by @Bekhouche in #3536
- fix(images): prevent loading on-demand upscalers by @fl0rianr in #4047
- fix(gui3): show CPU usage on macOS and KV cache usage on all platforms by @meghsat in #4069
- feat(gui): add image download and user message copy actions by @fl0rianr in #4062
- Add Phoronix Lemonade 2026.45.3 article to news page by @jeremyfowers in #4075
- feat(gui): add Halogen, Gufo, DwarfStar4 and TheNoise backend logos by @jeremyfowers in #4072
- Gate NVIDIA CUDA backends on a missing or outdated driver by @jeremyfowers in #4066
- feat(backends): Add description and author metadata to backends for display in the CLI and GUI by @danielholanda in #4059
- Update container backend image pins by @github-actions[bot] in #4083
- Add LFM3-1.7B and LFM3-VL-2B by @sawansri in #4044
- fix(gui): add multi GPU monitor by @fl0rianr in #4025
- feat(gui): add conversation image references and direct Vision LLM analysis by @fl0rianr in #4082
- feat(hrx): Enable MPT + add MTP capable model by @AaronStGeorge in #4014
- Beacon fixes: share the beacon port, escape the hostname by @festeraeb in #4094
- fix(test): Fix build for MSVC (Visual Studio 2026) by @julianxhokaxhiu in #4109
- fix(gui): pack backend cards into an aligned grid by @danielholanda in #4084
- Fix unhelpful API key error in the CLI by @MrAnayDongre in #4103
- test: remove network and cleanup flakiness from Ollama, Whisper and endpoint tests by @sreeram-11 in #4116
- fix(server): prevent auth bypass with an empty admin API key by @fl0rianr in #4122
- fix(models): list every shard in model file inventory by @pwilkin in #4108
- test: wait for DirectoryWatcher detection instead of fixed sleeps by @sreeram-11 in #4117
- Add GGUF zero-shot routers on llama.cpp: encoder in llama-server, rule head in lemond by @meghsat in #4096
- Fix rpm creation issues on OpenSUSE by @DigitalFlow in #4091
- ci: harden workflows against tzdata hangs, artifact upload errors and unprovisioned runners by @sreeram-11 in #4115
- fix(docs/api): describe optional image generation metadata by @fl0rianr in #4125
- fix(gui3): Upgrading banners to GUI3 by @meghsat in #4130
New Contributors
Full Changelog: v2026.45.3...candidate-v2026.47.0
The Windows installers in this release candidate are not signed; Windows SmartScreen will warn before installing. Stable releases are signed.
