Call Python functions

Modal’s own Python client deploys most Modal apps. Deploy the function with modal deploy as usual, and a Worker or Lambda calls it by app and function name, with arguments and results encoded as CBOR.

Bind the function by name

src/Api.ts

import * as Cloudflare from "@alchemy.run/cloudflare";
import * as Modal from "@alchemy.run/modal";
import * as Effect from "effect/Effect";
import * as HttpServerResponse from "effect/http/HttpServerResponse";

export default class Api extends Cloudflare.Worker<Api>()(
  "Api",
  { main: import.meta.url, compatibility: { flags: ["nodejs_compat"] } },
  Effect.gen(function* () {
    const embed = yield* Modal.InvokeFunction({ app: "inference", function: "embed" });
    return {
      fetch: Effect.gen(function* () {
        const vector = yield* embed.call<number[]>(["hello world"]);
        return yield* HttpServerResponse.json({ dims: vector.length });
      }).pipe(Effect.orDie),
    };
  }).pipe(Effect.provide(Modal.InvokeFunctionHttp)),
) {}

Arguments and results are CBOR: numbers, strings, bytes, lists, dicts, booleans and None.

Call a class method

const model = yield* Modal.InvokeFunction({ app: "inference", function: "Model.generate" });

Class.method names a method of a Modal class.

Start a call and read it later

const job = yield* train.spawn([], { epochs: 3 });
const metrics = yield* train.result(job, { timeout: "30 minutes" });

spawn returns as soon as Modal queues the call, so a long job doesn’t hold the request open.