Gemini 3.6 Flash
gemini-3.6-flash
Total Context
1M
Max Output
65.5K
Released
N/A
Input$1.5 / M tokens
Output$7.5 / M tokens
Gemini 4 Argon
gemini-4-argonGemini 4 Argon is Google's flagship reasoning model, built for complex software engineering, deep mathematical and scientific problem solving, and autonomous multi-step agents. Google trained it to keep context across long, multi-step tasks and to turn long documents mixing text, tables, charts and images into a single grounded output. It leads the Vals Index across finance, legal, coding and tax work, and supports long decode of up to 1M output tokens in a single response.
Context Window
1M tokens
Maximum Output
65.5K tokens
Release Date
Oct 13, 2026
Modalities
Gemini 4 Argon Pricing
| Input Price | Output Price | Cache Read | Cache Create 5m |
|---|---|---|---|
| $4/M | $20/M | $0.2/M | $0.375/M |
Gemini 4 Argon API Capabilities
Reasoning
Supported
Tool calling
Supported
Temperature parameter
Not supported
Attachments
Supported
Knowledge Base
2026-03-01
Endpoint Protocols
Messages APIgeminiCompletions API
Gemini 4 Argon Model Highlights
Deep reasoning, sustained engineering and multimodal understanding.
Deep Problem Solving
Built for difficult coding, mathematical and scientific problems, with adjustable reasoning depth.
Long-Horizon Engineering
Designed to maintain context through complex, multi-step software engineering and autonomous agent workflows.
Unified Million-Token Budget
Input, reasoning and output share 1,048,576 tokens; Google’s long-decode workflow extends responses beyond the standard 64K output limit.
Native Multimodal Understanding
Combines text, image, audio, video and PDF inputs to produce text-based analysis and synthesis.
Gemini 4 Argon Use Cases
Practical workloads for complex development, research and document analysis.
Codebase Modernization
Plan and implement multi-file refactors or legacy-code migrations, then review changes against project requirements.
Multimodal Document Synthesis
Turn long documents containing text, tables, charts and images into coherent reports grounded in the supplied material.
Scientific And Mathematical Analysis
Explore difficult technical questions, develop mathematical arguments and draft research analyses for independent verification.
Iterative Agent Projects
Build workflows that plan, create, inspect and revise software or 3D prototypes across multiple steps.
How to Use Gemini 4 Argon via the TokenHub API
curl 'https://us-api.tokenhub.com/v1/messages' \
-X 'POST' \
-H "Authorization: Bearer $TOKENHUB_API_KEY"Gemini 4 Argon Benchmarks
Gemini 4 Argon (High)
| Index score | ||
|---|---|---|
| Artificial Analysis Intelligence Index | Artificial Analysis broad capability aggregate | 52.6 |
Metrics sourced from Artificial Analysis
Gemini 4 Argon FAQs
Capabilities, practical tradeoffs and TokenHub integration guidance.
What Is Gemini 4 Argon?+
It is Google DeepMind’s flagship reasoning model for complex coding, autonomous agents, mathematics and science.
Which Workloads Suit Gemini 4 Argon?+
Use it for multi-file engineering, multimodal document synthesis and technical investigations that require sustained reasoning.
Which Inputs And Outputs Does Gemini 4 Argon Support?+
Google documents text, image, audio, video and PDF inputs with text output. Image and audio generation require other models.
What Limits And Tradeoffs Should I Consider?+
It can hallucinate, and difficult prompts may be slow. The 1M-token budget is shared; outputs beyond 64K need Google’s long-decode workflow and route support.
How Do I Use Gemini 4 Argon Through TokenHub?+
If listed for your account, select Gemini 4 Argon and use TokenHub’s displayed model ID, endpoint and API key. Confirm route-specific features before integrating.
Is Gemini 4 Argon Available, And How Is It Priced?+
Google offers paid API access. Check TokenHub’s current listing for pricing, routing and account availability; Google’s direct pricing may differ.
When Should I Choose Gemini 4 Argon Over A Lighter Model?+
Choose it when reasoning depth, large mixed-media inputs or extended engineering workflows matter. Compare accuracy, latency and total token cost on representative tasks.
Ready to use Gemini 4 Argon?
Use one API key to access Gemini 4 Argon and more AI models through TokenHub.
Gemini 4 Argon Media and Demos
Verified official posts and demos, alongside early community experiences.
X (Twitter)
Gemini 4 Argon App Rollout
Google Gemini ·
Official announcement introducing Gemini 4 Argon in the Gemini app.
The moment you've been waiting 4… Our new frontier model, Gemini 4 Argon, is rolling out now in the Gemini app to bring your most complex ideas into reality. See how Gemini 4 Argon and our latest updates across the Gemini app help you get the most out of every prompt. 🧵
View originalGemini 4 Argon From Plans To Deliverables
Google Gemini ·
Shows Gemini app workflows producing web apps, 3D files, slides and reports with Gemini 4 Argon.
Gemini 4 Argon breaks down your tasks step-by-step and instead of stopping at a text outline, Argon helps you build the real thing, whether that means: ➡️ Generating a shareable web app ➡️ Giving you ready-to-print 3D files ➡️ Producing slides and reports that are ready to go https://t.co/V4RPXun8I9
https://pbs.twimg.com/amplify_video_thumb/2110039533853212673/img/r4OrZZO4Lsf8FhNo.jpgView originalGemini 4 Argon Coding In The Gemini App
Google Gemini ·
Announces coding improvements and Antigravity’s coding interface in the Gemini app for macOS.
Gemini 4 Argon also brings major improvements in coding, so we’re making it easier to switch between chatting and coding with Google Antigravity's agentic coding interface now available directly in the Gemini app for macOS. https://t.co/waB3VmjmDI
https://pbs.twimg.com/amplify_video_thumb/2110039768579080193/img/q_yj8_IP7nVoDQHf.jpgView originalReddit
Gemini 4 Argon: Initial Coding Review
Gemini 4 Argon Initial review
this-is-hilarours ·
A community member compares backend implementation plans; replies discuss testing methodology and mixed coding experiences.
Read full post
As an ultra user just got the access to Gemini 4 Argon. i made an initial test in my python backend project . the project is already in production and there was a real feature integration that i have started working today . so i was curious how it would compare against Opus 5.5 . I have a claude code subscription and i also use a factory droid subscription . so i created an implementation plan based on the spec with opus 5.5 with extra high effort level . similarly created an implementation plan with Argon high . Both of the plan was fed to Gpt 6 astra who worked as judge as i believe gpt model always works as strong review model . i asked it to carefully judge and give both of them score out of 100 . here is the score astra give Opus 5.5 Score - 77/100 Gemini 4 Argon Scores - 59/100 I gave the original score and review to Gemini in antigravity it agreed with all of that . i think more testing would be needed from my side to test it properly . but it needs at least one more reasoning level like extra high One thing surprised me It compacted the conversation relatively faster . i don't know what is the context window set here but previously i never got the context compacted this fast . pretty sure the context is not 1 million current set in antigravity app
Gemini 4 Argon: Release Discussion
Gemini 4 Argon Release Blog
elemental-mind ·
Community thread sharing Google’s announcement of wider Gemini 4 Argon availability.
View originalGemini 4 Argon: Early User Feedback
Thanks for Gemini 4 Argon
Liberation2020 ·
An Ultra subscriber reports positive task results; the discussion covers speed, usage limits and personal workflows.
Read full post
I’m making this for the Google team that may be checking the thread. I’m an Ultra user and have patiently been waiting for this model. It is everything I need and more. I’ve been seeing others getting great results from it. As well. It’s ability to turn some of my things into skills that actually work is next level compared to what was happening before. It’s work rate is also impressive in my opinion detailed heavy task that I feel should maybe take a little while are done quite quickly. I’m still fine running my request style with it but I am quite happy with it this past couple of hours. It even did better on a task I had set to Astra to complete. (Not saying it is overall better just did that one task better) Bottom line I’m happy and I’m super excited to see nano banana pro update and also the video models!!!
YouTube
Gemini 4 Argon: Ten Practical Demos
Gemini 4 Argon: 10 incredible AI demos (1M token output in action)
Google Cloud Tech
Official showcase covering code migration, interactive 3D projects and multimodal deliverables.
View originalGemini 4 Argon: From Concept To 3D Cathedral
Gemini 4 Argon and Google Antigravity: From 2D concept to 3D cathedral
Google for Developers and Google Antigravity
An agent uses Gemini 4 Argon, Antigravity and Blender to reconstruct a cathedral from a 2D reference.
View originalGemini 4 Argon: Legacy Code To Rust
Gemini 4 Argon: Modernizing 40 year code
Google Antigravity and Google for Developers
Official demonstration of Antigravity and Gemini 4 Argon modernizing decades-old scientific code into Rust.
View original