For Developers/Models/Gemini 4 Argon
Closed SourceGoogleReleased 2026-09

Gemini 4 Argon

Google's frontier model for real-world coding, enterprise knowledge work, and cybersecurity defense

Context window

1M (shared input + output)

Input / 1M tokens

$2.00

Output / 1M tokens

$10.00

Provider

Google

Introductory pricing through November 30, 2026: $2/$10 per 1M input/output tokens (cached input $0.10). From December 1: $4/$20 (cached $0.20). Batch is half price. Paid API tier only; not on the free tier. Input, thinking and output share one 1,048,576-token window; synchronous responses default to a 65,536-token output cap, and long decode raises that to the full window. · Data verified 2026-11-17

Gemini 4 Argon is Google's frontier model built to sustain deep reasoning across complex, long-horizon workflows. It targets software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense, and can write up to 1 million output tokens with long decode. After a two-week restricted rollout to vetted cyber defenders, it became generally available to paid API customers on October 13, 2026. Gemini 4 Carbon became generally available on November 10, 2026 and is now Google's most capable model, and Gemini 4 Hydrogen (November 12) is the faster, cheaper Gemini 4 model at $1.50/$7.50, which Google reports edges Argon on the Artificial Analysis Intelligence Index (51 vs 50). Google positions Argon between them, for the majority of advanced long-horizon tasks with lower token use than Carbon.

Capability index

Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.

Coding
90
Reasoning
95
Math
85
Multimodal
75
Long context
100
Speed
70
Cost efficiency
85

How to access it

Generally available since October 13, 2026 on the paid Gemini API tier (model ID gemini-4-argon), and in the Gemini app for Google AI Ultra and Pro subscribers.

Strengths

  • ✓Long output context (1M tokens with long decode)
  • ✓Strong performance on software engineering tasks
  • ✓Lower token use per turn than Gemini 4 Carbon, per Google
  • ✓Advanced reasoning capabilities

Best for developers who...

Long-form software engineering tasksCybersecurity vulnerability analysisEnterprise knowledge work (legal, finance)Complex multi-step problem solving

When to choose it (and when not to)

Reach for Gemini 4 Argon when...

  • →Projects requiring extended output sequences
  • →Deep reasoning on complex workflows
  • →Frontier work before December 1, 2026, while the introductory $2/$10 price applies

Look elsewhere if...

  • ✕Everyday coding and agent tasks: Gemini 4 Hydrogen costs $1.50/$7.50, has a free tier and scores 51 vs 50 on the Artificial Analysis Intelligence Index, per Google
  • ✕Cost-sensitive work after December 1, 2026, when the price doubles to $4/$20
  • ✕You need the free API tier (Argon is paid tier only)

How to use it

Quickstart

Python
from google import genai

client = genai.Client()

interaction = client.interactions.create(
    model="gemini-4-argon",
    input="Review this function for security issues: ...",
)
print(interaction.output_text)

Install with pip install -U google-genai. Pattern from Google's Gemini API quickstart (Interactions API); needs a paid-tier API key.

API model id: gemini-4-argon

Benchmarks

BenchmarkScoreNotes
DeepSWE v1.177.9% (highest thinking)Google model card, mini-swe-agent harness
LVBench91.7%Long video understanding benchmark
Vals Index68.9%General AI capability benchmark
AutomationBench (Zapier)51.3%Google-reported, #1 at launch
CWE-bench v168%Google-reported, tied for first
Artificial Analysis Intelligence Index v5.0.150Per Google's Gemini 4 Hydrogen announcement and Artificial Analysis (November 2026); Gemini 4 Hydrogen scores 51 at its default high level
Terminal-Bench 4.057.4% (highest thinking)Google model card
OSWorld 2.0 offline subset69.2%Google model card, partial score, best of 3 runs

Source: Gemini 4 Argon model card

Compare Gemini 4 Argon

Compare Gemini 4 Argon with any other model

Build a comparison →
All model comparisons →

Learn the concepts