Ollama 0.40.16 corrected routing for its Claude Code launcher so the integration uses the model selected in Ollama. It also changed the Responses API compatibility layer to honor the request’s truncation setting.
Ollama
Also known as: Ollama CLI · Ollama API
Plain English
Ollama lets users download and run models through a command line or an app. Developers can also send requests to its API from their own software.
Technical Overview
Ollama packages model weights and settings and manages the runner that loads them. Its inference stack includes llama.cpp and a multimodal engine for supported models. The API offers streaming responses and tool-call messages when the selected model supports them.
Chronological Entries (3)
Ollama added support for tool calls in streamed responses. A client can display text as it arrives and process structured tool-call messages from the same response stream.
Ollama introduced its new multimodal engine. Image handling includes caching processed images, allowing the runtime to reuse work when a conversation repeatedly refers to the same picture.
Direct Relations (2)
- Part ofInference Engines
- Usesllama.cpp