Files
veridian/server/utils/ai-provider.ts
T
zoeissleeping 59bb7fbc12 Performance enhancements galore! New themining system
This is once again a huge commit, but its mostly performance
improvements along with some bug fixes and refactoring. It also includes
changes to the theming systems. I'm still not 100% happy with the
theming system, but its better than before.

Model fetching has been dramatically improved! Nearly all the important
computation and pre-processing has been moved to the server. This has
also somehow fixed the way model details are loaded, which was causing
many models to be missing their details despite models.dev having them.

The markdown renderer has once again been changed, but I'm mostly
certain that this is the last time major changes will be made to it. The
renderer is not spamming components, bloating memory usage, and its not
using a bug prone custom written chunking system.

There's also a lot more that I haven't mentioned and honestly forgot. I
need to get better commit hygiene tbh.
2026-02-20 00:20:50 -06:00

152 lines
5.5 KiB
TypeScript

import { createCerebras, type CerebrasProvider } from "@ai-sdk/cerebras";
import { createGoogleGenerativeAI, type GoogleGenerativeAIProvider } from "@ai-sdk/google";
import { createOpenRouter, type OpenRouterProvider } from "@openrouter/ai-sdk-provider";
import { createOllama, type OllamaProvider } from "ai-sdk-ollama";
import { createOpenAICompatible, type OpenAICompatibleProvider } from '@ai-sdk/openai-compatible';
import { createCohere, type CohereProvider } from '@ai-sdk/cohere';
import { type Entity } from "@triplit/client";
import { schema } from "~~/triplit/schema";
import { type StreamTextTransform } from "ai";
import { transformCerebrasReasoningStream } from "./cerebras";
import { providerBaseUrls } from "~/types/model";
import { type Result, Err, Ok } from "~~/types/result";
// import { createLongcatTransformer } from "./longcat";
export type ModelGateway = OpenRouterProvider | OllamaProvider | CerebrasProvider | GoogleGenerativeAIProvider | OpenAICompatibleProvider | CohereProvider;
export interface Gateway {
gateway: ModelGateway;
streamTransformer: StreamTextTransform<{}> | StreamTextTransform<{}>[] | undefined;
textTransformer: ((text: string) => string) | ((text: string) => string)[] | undefined;
}
export interface Provider {
gateway: Gateway | null;
endpoint: {
baseURL: string;
modelsEndpoint: string | null;
headers: Record<string, string>;
}
}
export enum GatewayFetchError {
NoProviderApiKey = 0,
NoProviderBaseUrl,
}
export async function getProviderDetails(provider: Entity<typeof schema, 'providers'>, providerApiKey?: string, model?: Entity<typeof schema, 'models'>): Promise<Result<Provider, GatewayFetchError>> {
let gateway: Gateway = {} as Gateway;
let baseURL = undefined;
let modelsEndpoint = undefined;
let headers: Record<string, string> = {};
if (provider.config.apiProxyUrl && provider.config.apiProxyUrl.trim() !== '') {
baseURL = provider.config.apiProxyUrl;
} else {
baseURL = providerBaseUrls[provider.type];
}
baseURL = baseURL.replace(/\/$/, '');
switch (provider.type) {
case 'openrouter': {
if (providerApiKey === undefined) {
return Err(GatewayFetchError.NoProviderApiKey);
}
gateway.gateway = createOpenRouter({
apiKey: providerApiKey,
headers: {
'HTTP-Referer': 'https://localhost:3000',
'X-Title': 'Veridian',
},
});
headers['Authorization'] = `Bearer ${providerApiKey}`
modelsEndpoint = `/models`;
} break;
case 'ollama': {
if (baseURL === undefined) {
return Err(GatewayFetchError.NoProviderBaseUrl);
}
if (providerApiKey !== undefined) {
headers['Authorization'] = `Bearer ${providerApiKey}`
}
modelsEndpoint = `/api/tags`;
if (model !== undefined) {
const innerGateway = createOllama({
apiKey: providerApiKey,
baseURL,
})
gateway.gateway = ((modelId: string) => innerGateway(modelId, { think: [...model.attributes.capabilities].includes('reasoning') })) as OllamaProvider;
}
} break;
case 'cerebras': {
if (providerApiKey === undefined) {
return Err(GatewayFetchError.NoProviderApiKey);
}
gateway.gateway = createCerebras({
apiKey: providerApiKey,
baseURL,
})
gateway.streamTransformer = transformCerebrasReasoningStream() as StreamTextTransform<{}>;
gateway.textTransformer = (text: string) => {
return text.split('</think>').at(-1)!.trim()
};
headers['Authorization'] = `Bearer ${providerApiKey}`
modelsEndpoint = `/models`;
} break;
case 'google': {
if (providerApiKey === undefined) {
return Err(GatewayFetchError.NoProviderApiKey);
}
gateway.gateway = createGoogleGenerativeAI({
apiKey: providerApiKey,
baseURL,
});
headers['x-goog-api-key'] = `${providerApiKey}`
modelsEndpoint = `/models`;
} break;
case 'longcat': {
if (providerApiKey === undefined) {
return Err(GatewayFetchError.NoProviderApiKey);
}
gateway.gateway = createOpenAICompatible({
name: 'LongCat',
apiKey: providerApiKey,
baseURL: baseURL ?? providerBaseUrls[provider.type],
includeUsage: true,
})
headers['Authorization'] = `Bearer ${providerApiKey}`
modelsEndpoint = null;
// streamTransformer = createLongcatTransformer() as StreamTextTransform<{}>;
} break;
case 'cohere': {
if (providerApiKey === undefined) {
return Err(GatewayFetchError.NoProviderApiKey);
}
gateway.gateway = createCohere({
apiKey: providerApiKey,
baseURL,
});
headers['Authorization'] = `Bearer ${providerApiKey}`
modelsEndpoint = `/models`;
} break;
}
return Ok({
gateway,
endpoint: {
baseURL,
modelsEndpoint,
headers,
}
});
}