Metadata-Version: 2.4
Name: wave-dispatch
Version: 0.4.3
Summary: wave Dispatch — local-first AI router. `dispatch serve` runs an OpenAI-compatible proxy that routes each request to the cheapest capable model (local-first; escalate to your frontier only when needed). BYO keys + infra.
Author: WAVE Online, LLC
License: MIT
Project-URL: Homepage, https://dispatch.wave.online
Project-URL: Repository, https://github.com/wave-av/dispatch-edge
Keywords: llm,routing,local-first,cost,edge,dispatch,agents,x402
Requires-Python: >=3.8
Description-Content-Type: text/markdown

# wave-dispatch (Python)

Thin client for [wave Dispatch](https://dispatch.wave.online) — route each request to the cheapest
capable model (local-first; escalate to your frontier only when needed). BYO keys + infra. Zero deps.

```python
from wave_dispatch import Dispatch

d = Dispatch()                      # reads WAVE_LICENSE env; omit license for x402 pay-per-use
print(d.route("find the auth handler"))   # {route, probability, margin, forward}
# d.execute("name 3 colors")        # run on the edge (if your plan allows)
# d.route_vector(vec768)            # matmul-only: cheapest/fastest
```

Python 3.8+. The edge worker is open-source: github.com/wave-av/dispatch-edge. A WAVE product.
