Reflection: Beam 1.0 501B A23B reflection/beam-1.0-501b-a23b

О модели Reflection: Beam 1.0 501B A23B

Beam 1.0 is Reflection's open-weight reasoning model for coding and long-horizon agentic workflows. It uses a sparse mixture-of-experts architecture with 501B total and 23B active parameters per token, and a...

Использование Reflection: Beam 1.0 501B A23B через НейроГейт

OpenAI-совместимый эндпоинт https://api.neuralgate.ru/v1 — замени base_url в любом SDK и используй модель по её id.

# Chat completion
curl https://api.neuralgate.ru/v1/chat/completions \
  -H "Authorization: Bearer $NEURALGATE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "reflection/beam-1.0-501b-a23b",
    "messages": [{"role":"user","content":"Привет!"}]
  }'

from openai import OpenAI

client = OpenAI(
    api_key="sk-...your-neuralgate-key...",
    base_url="https://api.neuralgate.ru/v1",
)
response = client.chat.completions.create(
    model="reflection/beam-1.0-501b-a23b",
    messages=[{"role": "user", "content": "Привет!"}],
)
print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.NEURALGATE_API_KEY,
  baseURL: "https://api.neuralgate.ru/v1",
});
const resp = await client.chat.completions.create({
  model: "reflection/beam-1.0-501b-a23b",
  messages: [{ role: "user", content: "Привет!" }],
});
console.log(resp.choices[0].message.content);

Похожие модели

Сравнить Reflection: Beam 1.0 501B A23B с другими

Кейсы использования

Где Reflection: Beam 1.0 501B A23B часто применяется: