Run 35B MoE models in ~2.5 GB of RAM. Weights stream from SSD; a trained router predicts loads a step ahead. Mac, iPhone, Android, Windows supported
Python
// This file was generated by [ts-rs](https://github.com/Aleph-Alpha/ts-rs). Do not edit this file manually.
import type { Attachment } from "./Attachment";
import type { MessageStatus } from "./MessageStatus";
import type { Role } from "./Role";
import type { Usage } from "./Usage";
import type { JsonValue } from "./serde_json/JsonValue";
export type MessageRecord = { v: number, role: Role, content: string,
reasoning_content?: string | null, attachments?: Array<Attachment>, tool_calls?: Array<JsonValue> | null, usage?: Usage | null, status: MessageStatus,
code?: string | null, model: string,
params?: JsonValue | null,
/**
* epoch ms.
*/
created_at: number, thread_id?: string | null, } & ({ [key in string]: number | string | boolean | Array<JsonValue> | { [key in string]: JsonValue } | null });