luminet.ai
Back to catalog
DeepSeekHosted on Luminet

DeepSeek V4

deepseek/deepseek-v4

Try in Playground

1.2T-parameter MoE (52B active) succeeding V3.2. SOTA on SWE-Bench Verified among open models. FP8-native training, 256K context.

Context window
256Ktokens
Input price
$0.32/ 1M tokens
Output price
$1.28/ 1M tokens
P50 latency
95ms

Quick start

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.luminet.ai/v1",
  apiKey: process.env.LUMINET_API_KEY,
});

const resp = await client.chat.completions.create({
  model: "deepseek/deepseek-v4",
  messages: [{ role: "user", content: "Hello!" }],
});

Capabilities

textcode

Throughput: 480 tok/s · Released 2026-04