Skip to main content
Dispatch lets you mix and match models per task. Use Claude for deep reasoning, GPT for broad generation, Gemini for speed — all through one interface.

​
Using Different Models

Reference any model by name in your dispatch command: 
/dispatch use opus to review this PR for edge cases

​
Model Selection Logic

Dispatch follows a clear resolution order:

​
1. Explicit Model Mentioned

If you mention a model name in your prompt, Dispatch uses it: 
/dispatch use sonnet to find better design patterns
This explicitly uses Claude Sonnet.

​
2. Multiple Models Mentioned

If multiple models are named, Dispatch uses the last one mentioned: 
/dispatch have opus review and sonnet test the auth module
This dispatches a single task using sonnet (the last model mentioned).
For truly independent tasks with different models, dispatch them separately:
/dispatch use opus to review the auth module
/dispatch use sonnet to test the auth module

​
3. No Model Mentioned

If no model is mentioned, Dispatch confirms your default before proceeding: 
User: /dispatch do a security review

Dispatch: I'll dispatch this using **opus** (your default). Sound good?
This prevents accidental use of expensive models.

​
4. Model Not in Config

If you reference a model not in your config, Dispatch auto-discovers it:
  1. Runs agent models to check availability
  2. Auto-adds it to config with the appropriate backend
  3. Dispatches immediately
/dispatch use claude-4.7-sonnet to review this code

Found claude-4.7-sonnet via agent models. Added to config with claude backend.
Dispatching security-review using claude-4.7-sonnet...

​
Creating Aliases with Custom Prompts

Aliases are named shortcuts that combine a model with role-specific instructions.

​
Define an Alias

Edit ~/.dispatch/config.yaml or use natural language: 
/dispatch create a security-reviewer alias using opus
This generates: 
aliases:
  security-reviewer:
    model: opus
    prompt: >
      You are a security-focused reviewer. Prioritize OWASP Top 10
      vulnerabilities, auth flaws, and injection risks.

​
Use an Alias

/dispatch have security-reviewer check my latest changes
This:
  1. Resolves to the opus model
  2. Prepends the security-focused prompt to the worker’s instructions
  3. Dispatches the task

​
More Alias Examples

aliases:
  perf-reviewer:
    model: sonnet
    prompt: >
      You are a performance optimization expert. Focus on
      algorithmic complexity, database queries, and bundle size.

​
Auto-Discovery of New Models

When new models become available (e.g., Claude releases opus-5.0), Dispatch automatically handles them:
  1. You reference the new model in a dispatch command
  2. Dispatch runs agent models to verify availability
  3. If found, it’s auto-added to your config
  4. The task is dispatched immediately
No manual config editing required.

​
Example Flow

/dispatch use opus-5.0 to review this architecture

Dispatcher: opus-5.0 not in config. Checking availability...
Dispatcher: Found opus-5.0 via agent models. Adding to config with claude backend.
Dispatcher: Dispatched architecture-review using opus-5.0.

​
Model-Backend Mapping

Dispatch intelligently routes models to backends:
Detection: Model ID contains opus, sonnet, or haiku (including versioned variants like sonnet-4.6, opus-4.5-thinking)Backend: claude (when Claude Code CLI is available)Command: claude -p --dangerously-skip-permissions (no --model flag)Examples:
  • opus
  • sonnet
  • haiku
  • claude-opus-4-6
  • sonnet-4.5-thinking

​
Complex Multi-Model Example

Pre-launch sweep for a fullstack SaaS app using multiple models: 
/dispatch we launch Thursday, need a full pre-launch sweep:

1) security audit the auth flow (JWT, sessions, password reset) and the
   stripe payment integration. use opus, do it in a worktree

2) performance - load test the top 5 API endpoints, find N+1 queries,
   check db indexes on users/orders/products tables. sonnet in a worktree

3) frontend audit - accessibility (WCAG 2.1 AA), bundle size, flag any
   components missing error boundaries. dispatch sonnet

4) generate OpenAPI specs for every endpoint that's missing one. use gemini

5) error handling audit - find uncaught exceptions, missing try/catch,
   inconsistent error response formats across all API routes. sonnet, worktree

ask me if anything is unclear. security is highest priority - flag
anything critical immediately
Result: 
All 5 workers dispatched with their monitors. Here's the summary:

Security (worktree):
  ● security-audit — Opus 4-6 (worktree): Full auth + payment security review

Performance (worktree):
  ● perf-audit — Sonnet 4-6 (worktree): API load testing + DB index review

Frontend (single worker):
  ● frontend-audit — Sonnet 4-6: Accessibility, bundle size, error boundaries

Docs (single worker):
  ● api-docs — Gemini 2.5 Flash: Generate missing OpenAPI specs

Error handling (worktree):
  ● error-handling — Sonnet 4-6 (worktree): Uncaught exceptions + error format audit

All running in background. Five workers, three models, three worktrees —
your session wrote five checklists and is idle. Dispatch more work or say
"status" anytime.
See the Configuration guide for details on editing your config and the Parallel Tasks guide for managing multiple concurrent workers.

Build docs developers (and LLMs) love

Get started for freeTalk to us