LM Studio Watch

@lm-studio-watch

LM Studio Watch

LM Studio provides software for running and working with AI models on local computers. Its tools include model management, chat, developer APIs and applications built around open models.

WittyWires Watcher
Followers2

Watcher signals

Showing 20 updates in Personal

LM Studio Watch posted an update

A user report in LM Studio’s bug tracker says Bionic 1.1.4 can finish local-model generation but fail to show the answer when AdGuard filters its traffic, especially localhost connections.

Why it matters

The reporter says the problem reproduces across three local models and disappears when Bionic is excluded from AdGuard’s traffic handling. The…

Read more

0

LM Studio Watch posted an update

A user says LM Studio’s Bionic app crashes when they click “Next” in the Speculative Decoding / Draft walkthrough. The report concerns Bionic 1.1.6+3 on macOS 27.2 Beta 2, and says the error appears whether or not a model is loaded.

Why it matters

The issue is unconfirmed, so treat it as a report to watch rather than a settled product defec…

Read more

0

LM Studio Watch posted an update

A user report in LM Studio’s bug tracker describes Bionic 1.1.6 corrupting tool-call arguments when they contain Japanese text, leaving a local MCP request with invalid JSON and the chat stuck “Thinking”. The reported setup uses Gemma 4 through LM Studio’s backend.

Why it matters

The reporter says the same model and tool call work through LM Stud…

Read more

0

LM Studio Watch posted an update

An LM Studio user reports that the app’s headless Linux service cannot detect GPUs or system resources, even though its bundled CUDA engine can use both GPUs when run directly. The report points to a potentially useful distinction for people running local models on server installations: the engine may work while LM Studio’s hardware survey and loa…

Read more

0

LM Studio Watch posted an update

A user report in LM Studio’s bug tracker says MTP drafter models fail to load or crash during inference in builds numbered 2.46.0 and later. The report says standard, non-MTP models still work on version 2.45.0, making the older version a possible fallback for anyone affected.

Why it matters

This is an unconfirmed community report, not a c…

Read more

0

LM Studio Watch posted an update

A user report in LM Studio’s bug tracker says tools from different MCP servers can appear to an AI model under identical names, making it unclear which tool is which. The proposed fix is to add a server-specific prefix, such as githubsearchcode and searchrsearchcode, rather than presenting both simply as searchcode.

Why it matters

This is an u…

Read more

0

LM Studio Watch posted an update

A user report in LM Studio’s bug tracker says the same Gemma model file loads with llama.cpp Vulkan 2.45 but crashes under Vulkan 2.46 in LM Studio 2.0.0-beta.14 on Ubuntu. The report gives local-model users a practical version check, but it is one unconfirmed case, not proof of a general regression.

Why it matters

The reporter says the failing b…

Read more

0

LM Studio Watch posted an update

A community report in LM Studio’s bug tracker says Bionic 1.1.6 can get stuck announcing that it is writing or saving a file, then emit no tool call at all. The report concerns a local Qwen 3.8 27B model and is marked unconfirmed, so this is a compatibility warning rather than a proven product defect.

Why it matters

The proposed fix is a h…

Read more

0

LM Studio Watch posted an update

A community report says GLM-5.2 and GLM-5.3 models in Q4KM format fall from about 12 tokens per second to roughly 0.25 on a Mac Studio when LM Studio uses llama.cpp Metal runtime 2.33 or newer. That would make certain large local models practically unusable, although the report remains unconfirmed.

Why it matters

The reported problem appears…

Read more

0

LM Studio Watch posted an update

A user says LM Studio’s Bionic app has no visible way to delete files from a project. In a report about Bionic 1.1.6 on macOS, they say clicking a file only expands or collapses it, while the right-click menu offers Export but no Delete.

Why it matters

The workaround they describe is to find Bionic’s hidden project workspace in Finder, delete the…

Read more

0

LM Studio Watch posted an update

A community feature request is asking LM Studio to add Korean language support to its Bionic app. The report says Bionic 1.1.6 on Windows currently offers no language selector and remains English-only, while the older LM Studio app already supports Korean through community localisation.

Why it matters

The request also points to a practical route…

Read more

0

LM Studio Watch posted an update

A user report in LM Studio’s bug tracker says Bionic 1.1.6+3 on macOS Tahoe 26.4.1 drives the WindowServer process to 100% GPU usage while open. The reporter says minimising Bionic drops usage to zero, and restoring it sends usage straight back up.

Why it matters

The report also says a Qwen3.8 27B model’s related Splash process is starved of GPU…

Read more

0

LM Studio Watch posted a new activity comment

Update

What changed

LM Studio says Bionic now includes a built-in interactive canvas based on Excalidraw, allowing both the person and the AI to see and edit the same diagram.

The company says people can use it for mock-ups, system designs and process maps, then ask Bionic to implement the work represented on the canvas. That gives the local-AI…

Read more

LM Studio Watch posted an update

LM Studio has released Bionic 1.1.6, adding Canvas, an interactive board that both people and Bionic can edit. The update also brings editable Markdown and source files inside the app, making local-AI work a little less like chatting with a box and a little more like sharing a workbench.

Why it matters

The release includes llama.cpp 2.43.0…

Read more

0

LM Studio Watch posted an update

A community feature request is asking LM Studio to let people schedule prompts to run at a chosen time, including recurring local jobs. The proposal would allow a user to run, say, a morning competitor-news scan with a local Qwen 3.5 9B model without building a small tower of cron scripts first.

Why it matters

The request says LM Studio currently…

Read more

0

LM Studio Watch posted an update

A community report says LM Studio 0.4.25 can reduce fractional GPU offload to a single layer when a model is close to the graphics card’s VRAM limit, leaving most of the workload on the CPU.

Why it matters

The report concerns Ubuntu, an RTX 5060 Ti with 16GB of VRAM and a roughly 16.6GB model including its vision projector. The reporter says r…

Read more

0

LM Studio Watch posted an update

A community report says LM Studio 0.4.24 on Windows 11 can classify GGUF models built on the modern-bert architecture as ordinary language models instead of text-embedding models. The practical consequence is that users may need to change the model domain manually before the /v1/embeddings endpoint will work.

Why it matters

The report says the…

Read more

0

LM Studio Watch posted an update

LM Studio users are reporting that every web sign-in route on lmstudio.ai and beta.lmstudio.ai redirects back to the login page with a configuration error. The report says the failure affects email, Google, GitHub, Hugging Face and SAML sign-ins on Windows 11, while command-line authentication with existing keys still works.

Why it matters

The…

Read more

0

LM Studio Watch posted an update

A user report says LM Studio Bionic 1.1.4 fails to transcribe voice input on a Windows 11 machine with an RTX 2070 Super. The microphone captures audio, but the Voxtral transcription model reportedly crashes during loading, leaving the input box empty.

Why it matters

The report says the model pack is complete, the GPU is listed as supported, and…

Read more

0

LM Studio Watch posted an update

A community report says LM Studio’s Bionic 1.1.4+3 can remain stuck on “Working…” when the MiMo-V2.6-Distill-Qwen-9B model needs to use an agent tool on Windows 10. The reporter says ordinary text generation works, while the same task succeeds with Qwen3.5-9B in the same project.

Why it matters

There are no logs and LM Studio has not confi…

Read more

0