When to use incremental tool updates instead of legacy streaming?
If your Codex setup on Google Cloud Vertex AI relied on legacy batched tool updates, the new incremental default changes that behavior. One config.toml flag reverts to the old mode if you need it.
What it is
Codex now sends tool updates incrementally by default on Google Cloud Vertex AI, so updates stream as they happen instead of via the legacy behavior. Developers can opt out through a feature flag in config.toml.
Applied case
A developer running Codex against Google Cloud Vertex AI gets incremental tool updates automatically after this release, with no action required. If their setup depends on the legacy update behavior, they add the opt-out flag under [features] in config.toml. The release notes attribute the change to pull requests 60617 and 60848 in the Codex repo.
- Incremental tool updates on Vertex AI are now the default; one config flag reverts to legacy
When to use it
Use when you run Codex on Vertex AI and want tool updates streamed incrementally. Skip when a pipeline depends on the legacy update behavior.
[features]
incremental_tools_legacy_gcp = falseRead more
Recommended for you
- WorkflowVercel
When to use AI Gateway instead of direct image model APIs?
A separate MiniMax account, key, and bill just to generate product images slows you down until one gateway API key covers it.
- WorkflowVercel
How I filter agent tools without burning output tokens
Routing and policy checks with generative models spend tokens on text you never use.
- WorkflowVercel
When to use Muse Ember instead of a big reasoning model?
Agents repeat bounded sub-tasks like tool calls many times per run, and paying heavyweight-model prices for each one drains budget and latency.
- WorkflowVercel
When to use AI Gateway instead of calling model providers directly?
Scattered keys, per-provider billing, and outages on any single provider make direct model calls fragile; AI Gateway fixes this at no markup.
Enjoyed this article?
Subscribe for new articles. No spam. Unsubscribe anytime.