github lemonade-sdk/lemonade candidate-v2026.46.1
v2026.46.1 (release candidate)

pre-release2 hours ago

Headline

⚠️ These notes are AI generated and will be revised by a human ⚠️

  • Benchmark results can be submitted directly from lemonade bench and shared via claim link at https://bench.lemonade-server.ai.
  • Zero-shot classification is available via POST /v1/classify with a labels parameter, and as a zero_shot condition type in router policies.
  • TheNoise image generation gains a CUDA backend for Linux x86_64 alongside existing ROCm support.
  • New container backends for AMD Strix Halo (gfx1151) — including ds4:rocm, rocmfpx:rocm, gufo:rocm, and llamacpp:halobox — bring GPU-accelerated inference to the platform.
  • POST /v1/systemone forwards typed questions to a decision model, with support for router-classified choices and SystemOne zero-shot classification.

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; their values are now under recipe_options, so external clients reading the old fields will find them absent.
  • The /models response includes a user_configurable boolean flag on backend descriptor options; consumers that assumed all listed options are editable may show technical knobs like soft_pixel_cap, pcm_sample_rate, and pcm_channels to users.
  • FLM v1.0.11 (bumped from v1.0.7) returns max_tokens exactly as requested rather than max_tokens+1, so chat completions that relied on the extra token will now stop one token early.
  • FLM v1.0.11 streams token usage counts only when stream_options.include_usage is set; without it the server's /api/v1/stats endpoint records no token counts or throughput for FLM streams.
  • FLM v1.0.11 returns 400 for previously-accepted invalid fields including invalid tool_choice, forced tool calls on small Qwen3 models (up to 1.7B), conversation history on *-flash models, and requests with no messages or prompt.

Lemonade Server

Operating System Downloads
Windows lemonade.msi
Ubuntu 24.04+ Launchpad PPA
Debian 13 (x86_64) lemonade-server_2026.46.1-debian13_amd64.deb
Debian 13 (ARM64) lemonade-server_2026.46.1-debian13_arm64.deb
Fedora 43 (x86_64) lemonade-server-2026.46.1-fc43.x86_64.rpm
Fedora 43 (ARM64) lemonade-server-2026.46.1-fc43.aarch64.rpm
Fedora 44 (x86_64) lemonade-server-2026.46.1-fc44.x86_64.rpm
Fedora 44 (ARM64) lemonade-server-2026.46.1-fc44.aarch64.rpm
macOS Lemonade-2026.46.1-Darwin.pkg

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.

Platform Download
Ubuntu x64 lemonade-embeddable-2026.46.1-ubuntu-x64.tar.gz
Ubuntu arm64 lemonade-embeddable-2026.46.1-ubuntu-arm64.tar.gz
Windows x64 lemonade-embeddable-2026.46.1-windows-x64.zip
macOS arm64 lemonade-embeddable-2026.46.1-macos-arm64.tar.gz

What's Changed

Thanks @Zorro-Zhang, @bitgamma, @fl0rianr, @iswaryaalex, @jeremyfowers, @kenvandine, @meghsat, @mike2307, @pwilkin 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

New Contributors

Full Changelog: v2026.45.3...candidate-v2026.46.1


The Windows installers in this release candidate are not signed; Windows SmartScreen will warn before installing. Stable releases are signed.

Don't miss a new lemonade release

NewReleases is sending notifications on new releases.