---
last_updated: "2026-11-04T08:20:14.000Z"
title: "A smaller Tinybird Local image"
excerpt: "The Tinybird Local image is about a third smaller to download and uses less memory when idle."
date: "2026-11-06"
---

# A smaller Tinybird Local image

The Tinybird Local image is about a third smaller to download, so `tb local start` and CI jobs that pull it start sooner. It also uses about 380 MB less memory when idle.

Your local data, Tokens and volumes keep working after the update. See [Tinybird Local](/forward/core-concepts/tinybird-local#tune-memory-usage) to set the number of CSV workers or turn off services you don't use.

## Breaking changes

- **The Tinybird Local image no longer includes development tools**: The image now contains only what Tinybird Local needs to run. Tools such as `git`, `gcc`, `make`, `ssh`, `rsync`, `gpg` and `uv` are not in the container anymore. If your scripts run them inside the container with `docker exec`, run them on the host or in a separate container.
- **The `CLICKHOUSE_VERSION` environment variable is removed from Tinybird Local**: It reported 25.3.7.194, which is not the ClickHouse® version that Tinybird Local runs. Use `SELECT version()` to get the running version.

## Improvements and bug fixes

- [Forward]: Added the `TB_LOCAL_CSV_WORKERS` environment variable to set the number of CSV workers in Tinybird Local. Fewer workers use less memory.
- [Forward]: Added the `TB_LOCAL_MCP` and `TB_LOCAL_KAFKA` environment variables to turn off the MCP server and the data connector service for Kafka, DynamoDB and S3 in Tinybird Local. Turning off both saves about 360 MB of memory, for example in CI.
- [Forward]: Large sorts and aggregations in Tinybird Local spill to disk sooner.