Ollama v0.40.14 Adds Retry Logic for ChatGPT Disconnects and Improves Web Search Handling
Ollama's latest release introduces retryable ChatGPT disconnects and enhancements for cloud web search responses, along with other internal improvements.
What changed?
Ollama v0.40.14 introduces several updates. Most notably, ChatGPT disconnect failures within the app are now retryable, which improves reliability when interacting with ChatGPT. The server has been updated to decode cloud web search responses, enhancing how external search results are integrated. Additional internal changes include sorting long top-k rows in chunks for MLX, preserving readable response input content for OpenAI responses, and using model output limits for Claude Code.
Why does it matter to an everyday developer?
For developers building workflows or tools that rely on ChatGPT through Ollama, the new retry logic for disconnects reduces interruptions and improves the user experience by automatically handling temporary connection issues. The decoding of cloud web search responses means that developers can expect more consistent and accessible results from web-sourced queries, supporting better downstream processing or user presentation.
What can the developer do now?
Developers should upgrade to Ollama v0.40.14 to take advantage of improved reliability and web search integration. No code changes are required to benefit from the new retry logic and improved response handling, but verifying that workflows handle reconnects smoothly is a recommended best practice. Review the full changelog if your application depends on top-k row sorting, OpenAI response handling, or feature toggling for specific AI model output limits.
