Gemini 3.5 Flash API

Quickstart

This model uses the Gemini Developer API native format — the model id is in the URL path, not the request body. Compatible with the Gemini SDK (@google/generative-ai), Gemini CLI, and opencode by overriding the base URL.

bashcurl "https://api.skillboss.co/v1beta/models/gemini-3.5-flash:generateContent" \
  -H "x-goog-api-key: $SKILLBOSS_WHOLESALE_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [
      {"parts": [{"text": "Hello"}]}
    ]
  }'

Your first 200 response is the fastest way to confirm setup. From there, swap in your real prompt and tune the model-specific parameters listed below.

Authentication

Every request must include your wholesale key. The header name depends on the endpoint — match the SDK you're using:

bashx-goog-api-key: $SKILLBOSS_WHOLESALE_KEY

Gemini's native API uses x-goog-api-key (Gemini SDK + opencode default) or ?key= as a query parameter (Gemini CLI). Both work — pick whichever matches your client. Standard (non-wholesale) console keys are rejected at the gateway with 401.

Code examples

pythonimport os
from google import genai

client = genai.Client(
    api_key=os.environ["SKILLBOSS_WHOLESALE_KEY"],
    http_options={"base_url": "https://api.skillboss.co"},
)

resp = client.models.generate_content(
    model="gemini-3.5-flash",
    contents="Hello",
)
print(resp.text)

typescriptimport { GoogleGenerativeAI } from "@google/generative-ai";

// Point the Gemini SDK at SkillBoss by overriding the base URL.
const genAI = new GoogleGenerativeAI(process.env.SKILLBOSS_WHOLESALE_KEY, {
  baseUrl: "https://api.skillboss.co",
});

const model = genAI.getGenerativeModel({ model: "gemini-3.5-flash" });
const result = await model.generateContent("Hello");
console.log(result.response.text());

Video & image understanding

This model natively understands video and images, not just text. Add a fileData content part (a public video/image URL, or a YouTube link) alongside your text prompt and the model will summarize, transcribe, answer questions about, or pull timestamped moments out of it. Video is billed by input token (VIDEO modality) — roughly 5–10k tokens per minute of video. The first request on a new video can take 1–2 minutes while it is ingested; repeat requests on the same video hit the cache and return in seconds.

bashcurl "https://api.skillboss.co/v1beta/models/gemini-3.5-flash:generateContent" \
  -H "x-goog-api-key: $SKILLBOSS_WHOLESALE_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [
      {"parts": [
        {"text": "Summarize this video in 3 bullet points with timestamps."},
        {"fileData": {"fileUri": "https://www.youtube.com/watch?v=aqz-KE-bpKQ"}}
      ]}
    ]
  }'

For programmatic video editing, compose this understanding step with the Shotstack Edit timeline API: use Gemini to decide the cuts / captions / highlights, then render the result with a Shotstack timeline. No separate video-editing service required.

Parameters

Gemini Developer API parameters. The model id is in the URL path (not the body) — see the curl example above. Body fields documented here.

Endpoint

Errors

The API uses standard HTTP status codes:

Pricing