Ollama v0.15+ introduces a new 'ollama launch' command that sets up and runs coding tools like Claude Code, OpenCode, Codex, and Droid with local or cloud models, removing the need for environment variables or config files. Users can run a single command such as 'ollama launch claude' to select a model and start coding. Recommended local models include glm-4.7-flash, qwen3-coder, and gpt-oss:20b, while cloud options include glm-4.7:cloud, minimax-m2.1:cloud, gpt-oss:120b-cloud, and qwen3-coder:480b-cloud. Coding tools require a context length of at least 64000 tokens for best results. The update also extends the cloud service's coding session window to 5 hours with more generous usage limits, even on the free tier.
Table of contents
Get startedOne command setupRecommended models for codingExtended coding sessionsConfigure onlyQuestions this post answers
How do I set up Claude Code or OpenCode to use a local model through Ollama without editing config files?
Ollama v0.15+ includes a new 'ollama launch' command that automatically configures and runs coding tools like Claude Code, OpenCode, Codex, and Droid with local or cloud models, requiring no environment variables or config files. Running 'ollama launch claude' or 'ollama launch opencode' walks through selecting a model and launches the integration directly. daily.dev surfaces updates like this for developers wiring local models into their coding tools.
Which local Ollama models are recommended for coding with Claude Code or OpenCode?
Recommended local models include glm-4.7-flash, qwen3-coder, and gpt-oss:20b, with glm-4.7-flash requiring about 23 GB VRAM at a 64000 token context length. For cloud use, glm-4.7:cloud, minimax-m2.1:cloud, gpt-oss:120b-cloud, and qwen3-coder:480b-cloud are suggested, since coding tools perform best with full context length. Developers picking a coding model can track releases like this on daily.dev.
What context length should I set in Ollama for coding tools like Claude Code to work well?
Coding tools need a context length of at least 64000 tokens to work best, which must be updated in Ollama's settings since it is not the default. Insufficient context length can degrade the performance of integrations like Claude Code, OpenCode, and Codex when using Ollama-hosted models. daily.dev helps developers stay on top of configuration details like this before they hit a wall.