read_file tool call not available with local model gemma4:e4b running with Ollama

Question

I am running opencode with locally running gemma4:e4b with Ollama. I tried to ask it to read a file and it was not able to. It gave the message I am attaching as an image to this. The read tool from openCode is not used since the model looks for read_file tool. When I switch to openAI's API call after giving my token, it starts to work. Is there a separate config needed for LLMs running on local with Ollama to resolve tools calls?

This issue might be a duplicate of existing issues. Please check:

Gemma 4 (e4b) tool calling fails via Ollama OpenAI-compatible API — streaming tool_calls not recognized #20995: Same model (gemma4:e4b) via Ollama — tool calls not recognized/executed, with detailed root cause analysis and reproduction steps
gemma-4-26b and gemma-4-31b opencode interaction issues leading to tool loops/failures #21034: Same gemma-4 model family — tool call failures including file read operations via local inference
Local Ollama subagents return tool calls as text instead of executing them in task/delegate #21181: Local Ollama models returning tool calls as text instead of executing them

These issues describe the same underlying problem: gemma4 tool calling not working correctly through OpenCode's Ollama integration. Please check if any of these cover your case and consider adding your findings there instead.

You perhaps need to increase the context window in your Ollama settings (that's my assumption as it is saying it didn't get any user query but only a list of tools which was probably incomplete due to the small context window).

Once that is done, you'll need a stronger model to get anything done and perhaps use something else than Ollama.

there is also issue 16531 that blocks gemma4 support

Any solution for this issue?

@rekram1-node

@rekram1-node Hi, please, I also have this issue, any idea? I am using a Macbook Pro M4 base + gemma4:e4b, Ollama, I have the exact same error resposne

I have the same issue with Gemma4 E4B IT with a context window of 32k running with llama.cpp. Can share this "thinking" output from within Opencode

Steps:
  ┃  1. Read config.yml.
  ┃  2. Analyze the file contents, specifically around lines 24 and 33, to identify how the cluster/environment settings are missing or
  ┃  misconfigured.
  ┃  3. Formulate a plan and/or ask the user for clarification based on the findings.
  ┃
  ┃  I will start by reading the file.

     [read]:config.yml

use journalctl -u ollama -f to see if prompt > num_ctx
model's num_ctx (context window size) can be checked by running ollama ps

How to fix :
ollama run gemma4-31b

/set parameter num_ctx 65536
/save gemma4-31b-ctx64k
/exit

edit ~/.config/opencode/opencode.json to rename model to the newly saved model (gemma4-31b-ctx64k)
rerun opencode

Automated: closing question issues.

GitHub Issues are for bug reports and feature requests only — questions / support are no longer tracked here.

For help & questions: https://discord.gg/opencode
Billing issues with Opencode Go / Zen: help@anoma.ly

If this is a bug or feature request, please open a new issue with the appropriate template.

Navigation Menu

Repository navigation

read_file tool call not available with local model gemma4:e4b running with Ollama #21354

Question

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

read_file tool call not available with local model gemma4:e4b running with Ollama #21354

Description

Question

Activity

github-actions commented on Apr 7, 2026

justfortheloveof commented on Apr 7, 2026

BrutchsamaJeanLouis commented on Apr 10, 2026

leonelmorenoa commented on Apr 13, 2026

leonelmorenoa commented on Apr 13, 2026

srggrs commented on May 20, 2026

rkuo2000 commented on May 31, 2026

rekram1-node commented on Jun 23, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions