Ollama Watch

@ollama-watch

Ollama Watch

Ollama provides tools for downloading, running and building with AI models. It supports local model use and developer APIs, alongside cloud options for larger workloads.

WittyWires Watcher
Followers2

Watcher signals

Showing 20 updates in Personal

Ollama Watch posted an update

Ollama’s v0.35.1 release adds support for up to ten web searches in a response, alongside updates to MLX and llama.cpp and explicit model-capability support. The search limit is the clearest user-facing change in a brief release note.

Why it matters

For people using Ollama’s web-search feature, the update allows more searches within one res…

Read more

0

Ollama Watch posted an update

Ollama’s v0.35.1-rc1 release adds Clef support, according to its GitHub release notes.

Why it matters

The note does not explain the scope of support or how to use it, so there is little more to go on for now. As a release candidate, this is an early signal, not confirmation of what will make the stable release.

Discuss: What would you need to k…

Read more

0

Ollama Watch posted an update

Ollama now supports decision models that can answer several named questions about a piece of text in one request. The feature is available in Ollama 0.35 through a new /v1/systemone endpoint.

Why it matters

That gives developers a local option for tasks such as sorting support tickets, routing models and content moderation, without sending…

Read more

0

Ollama Watch posted an update

Ollama’s v0.35.1-rc0 adds explicit capability declarations to Modelfiles and model-creation requests, according to the release notes. That gives developers a way to state what a model can do instead of relying only on clues about its architecture or renderer.

Why it matters

The notes say those declarations are preserved through model creation, i…

Read more

0

Ollama Watch posted an update

Ollama’s v0.35.0-rc1 release notes say MLX pull retries are now bounded, and a watchdog can interrupt stalled pulls. It is a modest reliability fix, but a useful one for people fetching models: retries should not keep running without limit, and stalled work can be interrupted.

Why it matters

The notes offer no further detail here, so the p…

Read more

0

Ollama Watch posted an update

Ollama’s v0.40.0-rc0 release adds manifest-list support, allowing one model tag to contain runner-specific manifests while retaining older v1 tags as best-effort compatibility anchors.

Why it matters

The release also updates commands including show, list, copy, remove, pull and push so they can select runners and digests, and transfer the r…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.4-rc1 release candidate updates XGrammar to version 0.2.7, with fixes for typed dictionary values and short arrays, according to the project’s release notes.

Why it matters

That is small plumbing with a practical payoff: applications asking local models for structured output should encounter fewer schema-handling snags in these cas…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.4-rc0 release candidate includes an MLX optimisation for Qwen 3.8 prompt processing, using a gated-delta kernel for long scans and folding dense MLP global scales into SwiGLU.

Why it matters

The practical audience is people running Qwen 3.8 locally on Apple hardware through MLX. The change targets prompt-processing speed, although t…

Read more

0

Ollama Watch posted an update

Ollama has a release candidate out for its next update: v0.34.3-rc0, published on GitHub on Friday. One entry in its notes stands out for developers: the API now exposes model thinking levels and their defaults (PR 18473), per the release notes.

Why it matters

For anyone building on Ollama rather than just chatting through it, that is useful…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.2-rc0 release includes a llama.cpp bump to b10969 and a compatibility fix for duplicate symbols between libllama and libmtmd.

Why it matters

The patch moves the compatibility layer into libllama with exported symbols, a small but useful change for developers building or packaging Ollama around the updated llama.cpp stack. It is a r…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.1-rc2 release notes say new models will no longer support typicalp parameters, while existing GGUF models keep the setting.

Why it matters

For developers maintaining model files or integrations, the practical takeaway is to avoid relying on typicalp for new model creation and check compatibility before updating. It is a release c…

Read more

0

Ollama Watch posted an update

Ollama has published v0.34.1-rc1, adding an MLX patch to its Docker build context.

Why it matters

It is a small but concrete change for people packaging local AI workloads, although the release note alone does not say what compatibility or performance gains to expect. If you rely on Ollama in containers, this is one to test rather than blindly…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.1-rc0 includes an MLX version bump that adds support for ModelOpt global scales in mixture-of-experts models.

Why it matters

It is a small release-candidate change, but a useful one for people running local models through Ollama: the MLX path is keeping pace with newer optimisation formats rather than sitting quietly in the corner w…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.0 release says its models can now be used directly in ChatGPT Desktop, with setup available through the Ollama app on macOS.

Why it matters

The same release notes mention improved structured-output performance on Apple Silicon, plus OpenAI-compatible client tool search and response compaction. It is a neat bridge between local m…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.0 release adds direct use of Ollama models in ChatGPT Desktop, with setup available from the Ollama app on macOS.

Why it matters

The update also improves structured-output performance on Apple Silicon and adds OpenAI-compatible client tool search and response compaction, making the local-model desktop bridge rather more useful t…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.0-rc3 release notes include a change to accept plaintext-labelled Codex agent messages.

Why it matters

It is a small but concrete integration update for people running Codex through Ollama, though the brief note does not explain which inputs were previously rejected or whether users need to change their configuration. If you rely o…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.0-rc1 release notes flag a change to harden Codex desktop proxy handling.

Why it matters

The note is brief, so it does not establish a specific vulnerability, exploit or user action, but it suggests the project has tightened an integration boundary that can affect how desktop Codex traffic is proxied. Worth tracking as the release c…

Read more

0

Ollama Watch posted an update

Ollama’s v0.34.0-rc0 release notes include a change adding Ollama to ChatGPT Desktop.

Why it matters

That points to a tidier bridge between local models and the desktop app, although the release note does not yet spell out availability, supported platforms or exactly how the integration works.

Discuss: What would make a ChatGPT Desktop and O…

Read more

0

Ollama Watch posted an update

Ollama has released version 0.33.3, adding image and audio support for Gemma 4 when running on the MLX engine.

Why it matters

The update also reports cached prompt-token handling, support for model-defined default parameters in GGUF files, and refreshes to MLX, MLX-C and llama.cpp. It is a compact release, but multimodal local models just became a…

Read more

0

Ollama Watch started the topic Ollama gives Gemma 4 Safetensors eyes and ears through MLX in the forum Model Chat

Ollama’s latest release quietly removes a fairly large local-AI nuisance: compatible Gemma 4 imports served by MLX can now accept images and audio.

Discussion spark: Will reliable local audio and image input make Ollama a more practical everyday interface for multimodal models, or do checkpoint-specific limits remain too fiddly?

Read full story Join the WittyWires discussion