Gemini 4 Argon

Name: Gemini 4 Argon model facts
Creator: OSMEXO

google/gemini-4-argon

Google
Family · Gemini 4 Argon
Released 30 Sep 2026
Audio · File · Image · Text · Video → Text
Reasoning · adjustable
Tools · JSON · Vision

Not on OSMEXO yet

Request this model

Spec & list price

Context1M

Max output944K

Input / 1M$2.00

Output / 1M$10.005× input

Scores · standing

Intelligence52.6#9/489

LiveBench79.3#6/32

Usage

Tokens / day142.8B▲ 20% 7d

Usage rank#28

Requests / day2.2M

Tool calls / req0.32

Per request · 30 days

Avg input51.2K3.4× median

Avg output1.1K1.6× median

Cache hit78%1.1× median

Reasoning share32%0.6× median

Tool-call failures0.9%0.5× median

Consumption

Peak 282.2B 30 OctLow 39.7B 13 OctRank today #28

News & trends

311 mentions this week ▲ 697%this model · videos, posts, articles · 30 days

Trending · 14 days

Frontier CS added benchmark for Gemini 4 Argon on Antigravity1 points
Gemini 4 Argon: The Model and Safety Cards2 points
Claude Opus 5.5 + GPT-6 Astra + Gemini 4 Argon for FREE42.9K views
Gemini Argon is a humorous model1.4K upvotes
Claude Opus 5.5 vs Gemini 4 Argon Make a Minecraft Mod58.2K views
Google Gemini 4 Argon shatters 59-year-old nuclear fusion conjecture in just a few hours | Humans stalled for 59 years, but AI found a way in a few hours1.4K upvotes
@GeminiApp on X1.1M views

News

Frontier CS added benchmark for Gemini 4 Argon on Antigravitytwitter.com · 27 Oct
Gemini 4 Argon: The Model and Safety Cardsthezvi.substack.com · 26 Oct
Claude Opus 5.5 and Claude Sonnet 5.5 are now available in Google AntigravityGoogle Cloud · 20 Oct
What the Gemini agent means for startupsGoogle Cloud · 15 Oct

Benchmarks

bar = share of the top score · Benchmark data: Artificial Analysis

Humanity's Last Exam#4 of 465

57.1%

SciCode#5 of 160

61.8%

Long-context reasoning#55 of 393

79.7%

Intelligence index#9 of 489

52.6

Terminal-Bench 4.0#6 of 152

57.1%

Speed · median across 6 providers, latest day

Output speed#237/33449 tok/s1.4× slower than median

Fastest provider85 tok/sbest of 6 providers

First token#3134.43 s4.3× longer than median

First answer token#7935 sthinking included

#2 cheapest of 8 with a similar score

Value

Intelligence Index

Blended price per 1M tokens (log)

LiveBench suites

Browser82.8

Coding89.7

Data Analysis85.2

Design73.7

Long Agentic61.2

Reasoning95.1

Research83.1

▲ Strongest: Reasoning, Coding. ▼ Weakest: Long Agentic (61).

Technical details

Reasoning: on · medium
Tokenizer: Gemini
Moderation: none

Reasoning effort

high
medium · default
low

Request parameters 12

tools
tool_choice
response_format
structured_outputs
reasoning
reasoning_effort
include_reasoning
temperature
top_p
max_tokens

+2 more

stop
seed

Interesting facts

7.4× the output room of similar models (943.7K against a 128.0K median).
Released 33 days ago.

When to use

Best for

Hardest reasoning and research problems
Browser-driving and long agentic workflows
Analysis of very long documents and videos
Multimodal work across audio, images and files

Think twice for

High-volume simple chat (cheaper options exist)
Self-hosting (closed weights)
Image or audio generation

Research papers

Same family