Gemini 4 Argon

Gemini 4 Argon is the first model of Google's Gemini 4 family, opened to developers on 13 October 2026, and Google calls it its latest frontier model. Google built it to hold context across long, multi-step tasks and keep going until the job is done: refactors and new features across many files, finance work such as ledger reconciliations and variance analysis, legal research with verified citations, and research that pulls many sources into one structured answer. It reads long documents that mix text, tables, charts and images, and it accepts video and audio as well. Google reports that it ties for first place on CWE-Bench v1, at 68%, for fixing security vulnerabilities, and that it is its most resilient model yet against indirect prompt injection. Through Thalam it has a 1M-token context and writes up to 64K tokens per answer, with reasoning, function calling and structured outputs. The price is $2.00 in and $10.00 out per 1M tokens until 30 November 2026, then $4.00 and $20.00 from 1 December, Google's standard rate for the model.

Quick start

Use any OpenAI-compatible client. Just change base_url and your key.

from openai import OpenAI

client = OpenAI(
    api_key="tl-xxxxxxxxxxxxxxxxxxxxxxxx",
    base_url="https://api.thalam.ai/v1",
)

response = client.chat.completions.create(
    model="google/gemini-4-argon",
    messages=[
        {"role": "user", "content": "Hello from Thalam!"}
    ],
)

print(response.choices[0].message.content)