Finally: Searchable Model Picker Makes OpenClaw Configuration Painless

If you've ever run openclaw configure and stared at an endless scroll of model names, you know the pain. With providers like OpenRouter offering 200+ models, finding the one you want felt like scrolling through a phone book in 2026.

OpenClaw v2026.2.17 fixes this with a searchable model picker.

The Problem

The /model picker allowlist prompt in openclaw configure used to be a simple list. You'd scroll... and scroll... and scroll. Want gpt-5.2-turbo? Hope your arrow key doesn't get tired. Looking for a specific Anthropic model? Good luck finding it among the Claude variants.

For users with multiple providers configured, the list could easily hit 300+ entries.

The Solution

Now when you hit the model selection prompt, you can simply type to filter. The filtering uses tokenized matching, which means:

• Type gpt-5.2 → shows all GPT-5.2 variants
• Type openai/ → filters to only OpenAI models
• Type claude sonnet → finds Claude Sonnet models
• Combine terms like gpt-5.2 openai/ for precise matches

This is the same fuzzy search behavior you'd expect from tools like fzf or VS Code's command palette.

How to Use It

Just run the configuration wizard:

openclaw configure

When you reach the model selection step, start typing instead of arrow-keying. The list filters in real-time.

This also works when setting up the /model picker allowlist - the list of models users can switch between using the /model command. If you're curating a specific set of models for your team, you can now quickly find and select the ones you want.

Why This Matters

Small quality-of-life improvements like this make a big difference in daily workflows. Configuration is something you hopefully don't do every day, but when you do, it should be frictionless. This change takes what was a 2-minute scroll-fest and turns it into a 5-second search.

Technical Details

The implementation (PR #19010, thanks @bjesuiter!) adds tokenized matching to the searchable-select component. Tokens are split by common delimiters (spaces, slashes, dashes), so partial matches work naturally across provider prefixes and model name components.

Reference: GitHub PR #19010

Running v2026.2.17? Try openclaw configure and enjoy the new searchable model picker. Running older? Time to update.