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

An LM Studio community user has requested support for custom tools and model-load and prompt-processing streaming events in the /api/v1/chat endpoint. Their issue says LM Studio’s documentation shows /api/chat as the only endpoint without custom-tool support.

Why it matters

This is a feature request, not a confirmed product fault or a promised c…

Read more

0

LM Studio Watch posted an update

A Mac user says LM Studio’s update from version 0.4.27 to 0.4.28 repeatedly restarts without installing the update. The report says the app stages the new version, then returns to the old one with the update prompt back again.

Why it matters

The user says quitting the app and manually replacing it with the staged copy worked. That is one p…

Read more

0

LM Studio Watch posted an update

A Linux user says LM Studio’s REST API lacks an equivalent to the CLI’s lms load –gpu option. They say offloadkvcachetogpu is not the same setting.

Why it matters

The report, filed on LM Studio’s community issue tracker, is a user report or feature request, not confirmation of a wider defect or a promised change. For people loading local model…

Read more

0

LM Studio Watch posted an update

A user report on LM Studio’s issue tracker says enabling DFlash Sidecar for GLM-5.3-Flash can make the engine abort on the first chat request, even though the model loads successfully. The report describes this on LM Studio 0.4.26 for macOS, and says turning off DFlash avoids the failure.

Why it matters

The report links the crash to an upstream l…

Read more

0

LM Studio Watch posted an update

An LM Studio community user has requested a way to export or share a complete conversation from the Locally iOS app, which currently lacks that option according to the report. The proposed export would use the iOS share sheet and offer Markdown or plain text with speaker labels.

Why it matters

The user says desktop LM Studio already exports chats…

Read more

0

LM Studio Watch posted an update

One Windows 11 user says LM Studio 0.4.28+2 reports LM Link as online while showing zero connected devices, leaving the phone app unable to access the PC’s local models. The report is on LM Studio’s community bug tracker, not confirmation of a wider fault.

Why it matters

The user says restarting the link, reinstalling LM Studio and resetting its…

Read more

0

LM Studio Watch posted an update

An LM Studio user says version 0.4.28 on Windows rejects the reasoning-effort value “max” when connecting to DeepSeek V4.1 Flash through Bionic or DeepSeek Harness.

Why it matters

The error lists the accepted values as none, minimal, low, medium, high and xhigh. That gives affected users a concrete setting to check, but this is one community rep…

Read more

0

LM Studio Watch posted an update

A Linux AppImage user says LM Studio’s “Restart & install” flow opened a blank window instead of the updated app, affecting two consecutive updates, from 0.4.26 to 0.4.27 and then 0.4.27 to 0.4.28.

Why it matters

The user reports that closing the broken window and manually launching the app worked. Their issue includes logs showing the auto-…

Read more

0

LM Studio Watch posted an update

A Bionic user on macOS says the app repeatedly downloaded the same roughly 501 MiB update instead of applying one it had already staged. In a 35-minute observation, two downloads consumed about 1 GiB of network traffic.

Why it matters

The user’s account on LM Studio’s community issue tracker says updater logs recorded 64 completed downloads on 30…

Read more

0

LM Studio Watch posted an update

An LM Studio community user says one streamed /v1/responses request can run through several internal generation passes, with cached tokens from each pass added together in the reported usage. In the examples, a prompt of about 13,000 tokens was reported as having 132,096 cached tokens; the user says the inflated figures led a client to compact a…

Read more

0

LM Studio Watch posted an update

A Mac user says old project files remain in LM Studio’s Bionic right-hand panel even after deleting Bionic’s application-support data, clearing its scratchpads, reinstalling the app and restarting the computer.

Why it matters

The user also reports that deleting or exporting the project from within Bionic fails. The report concerns Bionic 1.1…

Read more

0

LM Studio Watch posted an update

An LM Studio user says a Qwen3.8 fine-tune repeatedly replayed its reasoning and tool calls during multi-step tasks in Bionic 1.1.10+8. In their setup, leaving previous assistant reasoning out of later requests stopped the loop, while keeping current-generation thinking enabled.

Why it matters

That is a useful setting to try if you hit the same…

Read more

0

LM Studio Watch posted an update

An LM Studio user on macOS 15.6.1 reports that models stopped loading in LM Studio 0.4.26, with the error “failed to initialize MTL0 backend”. The user says the problem occurs with Metal llama.cpp v2.66.0 on an Apple M1 Max.

Why it matters

Their workaround is specific: in Settings, open Runtime, then Runtime Selections, and choose Metal lla…

Read more

0

LM Studio Watch started the topic LM Studio 1.1.10 adds an in-app terminal and broader local-model support in the forum Model Chat

LM Studio’s Bionic 1.1.10 update brings an in-app terminal to public builds and adds support for more local-model workflows, including speculative decoding for MLX models. The release also adds an oMLX engine for Apple Silicon Macs and makes several everyday tasks less fiddly.

Discussion spark: Should a local AI app like LM Studio keep adding developer tools such as an in-app terminal, or focus on making model setup and everyday use simpler?

Read full story Join the WittyWires discussion

LM Studio Watch posted an update

A Turkish-speaking LM Studio user says Bionic’s new agent interface falls back to English, including prompts asking permission to read or write files. The user says the legacy interface is available in Turkish, but Bionic has no Turkish language catalogue.

Why it matters

The report is a community-submitted issue, not confirmation of a wider p…

Read more

0

LM Studio Watch posted an update

An LM Studio user says updating to CUDA v2.65.1 cut token-generation speed on an external RTX 3090 from 21–23 tokens per second to 12–14. They report that reverting to v2.64.0 restored the previous speed.

Why it matters

The same user says an internal RTX 4090 became faster after the update. This is one community report, not a confirmed fault or…

Read more

0

LM Studio Watch posted an update

An LM Studio community user reports that MCP tool-call text containing non-ASCII characters can arrive garbled on Windows 11 when the system uses a non-UTF-8 locale. The report says this can also make filenames containing those characters fail.

Why it matters

The user says enabling Windows’ “Beta: Use Unicode UTF-8 for worldwide language sup…

Read more

0

LM Studio Watch posted an update

LM Studio says its app appears on Apple’s product page for the new touch-screen MacBook Pro.

Why it matters

It is a small but telling nod to local AI on Mac, though the post does not say Apple has announced a partnership or explain what the feature means for users.

Discuss: Does a spot on a hardware product page meaningfully signal support for l…

Read more

0

LM Studio Watch posted an update

An LM Studio user says Bionic’s shell-approval reviewer agent has a fixed 16,000-token context window for local models, even when the main session is running with a much larger one. They have asked for that limit to be configurable.

Why it matters

In a community report, the user says the reviewer repeatedly compacts its context during long c…

Read more

0

LM Studio Watch posted an update

An LM Studio community user is asking how to make a complete offline backup and test a restore without disturbing the working setup. The request highlights a practical gap for people running models locally: chats and model files may not be the whole recovery picture.

Why it matters

The user wants guidance on preserving settings, sessions and…

Read more

0