google / Chat

Gemini 4 Argon

google/gemini-4-argon

Google's flagship Gemini reasoning model for complex software engineering, deep maths and science, autonomous multi-step agents and very long documents. Accepts text, images, audio and video in a 1,048,576-token context window.

Best for: Long-horizon coding, agents and million-token documents

ShareLinkedIn X

Catalog status

Enabled

Research checked

2026-10-14

Price / performance

What it costs to get this result

Benchmark

DeepSWE v1.1 · Source-reported results, not a composite ranking. Prices use the average of TokenRex input and output rates per million tokens; actual spend depends on usage. Test configurations differ.

Gemini 4 Argon77.9% · 5 sourced models

Blended price / 1M tokens · log scale

● Current model● Other modelsDashed line: price/score frontier among plotted models

Compare sourced models on this benchmark

Benchmark results

19 reported results

Results are source-attributed, not TokenRex measurements. Different tests and thinking variants are not interchangeable.

19 of 19 results

Vals Index

Knowledge work · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

68.9%

AutomationBench

Knowledge work · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card (Score), October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

51.3%

Vals Finance Agent v2

Knowledge work · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

65.4%

Harvey's Legal Agent Benchmark

Knowledge work · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

19.6%

DeepSWE v1.1

Agentic coding · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

77.9%

FrontierSWE v2

Agentic coding · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

55%

Vibe Code Bench

Agentic coding · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

91.9%

Terminal-bench 4.0

Agentic coding · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

57.4%

PostTrainBench

ML engineering · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

45.3%

Terminal-Bench Science 0.1

Science and math · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

57.6%

LABBench 2

Science and math · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

88.8%

RiemannBench

Science and math · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

76%

GraphWalks BFS (up to 128K)

Long context · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card (F1), October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

99.7%

GraphWalks BFS (256K to 1M)

Long context · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card (F1), October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

84.2%

Agent's Last Exam

Computer use · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card (Pass rate), October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

39.5%

OSWorld-2.0 (offline subset)

Computer use · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card (Partial score), October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

69.2%

Chartography

Multimodal understanding · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

71.6%

LVBench

Multimodal understanding · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

91.7%

CWE-bench v1

Cybersecurity · Vendor-reported

Test conditions and source

Self-reported by Google in the Gemini 4 Argon model card, October 2026.

Tested model: google/gemini-4-argon

Google DeepMind · Gemini 4 Argon model card · Checked 2026-10-14

68%

Model specifications

Context window

1.05M

Published provider limit

Maximum output

65.54K

Published provider limit

Knowledge cutoff

March 2026

Upstream model

Input and output modalities

Provider-reported formats; API compatibility is detailed below.

Input

Text

Image

Audio

Video

Output

Text

Exact model ID: google/gemini-4-argon
Model type: Chat
Context window: 1,048,576 tokens
Maximum output: 65,536 tokens
Knowledge cutoff: March 2026

A context window is the total conversation budget, not a separate maximum input allowance; generated output and reasoning can use that budget. Published provider limits are not independently tested TokenRex request limits.

Versions and thinking levels

One enabled version is currently listed for this model.

Gemini 4 ArgonSelected$2.00 in / $10.00 out per 1M tokens

Features and tool support

TokenRex’s exposed interface, not every feature advertised by the provider. “Available” describes the implemented interface, not a successful test of every input or tool.

Reasoning configuration

No independent reasoning-mode guarantee is recorded for this exact model.

See enabled versions and provider sources for supported controls.

Coding and agent workflows

Text/code generation and customer-managed tool loops.

Coding benchmark results do not establish a hosted terminal, sandbox, autonomous browser or guaranteed task success.

Streaming answers

Available

Incremental text through TokenRex’s chat endpoint; Claude also has native Messages access.

Customer-defined function tools

Limited

Only customer-defined function tools are forwarded. Model/protocol compatibility applies; your application authorizes and executes every tool.

Forced or named tool selection

Limited

The chat route accepts auto, none and required. A named-tool choice is not forwarded; exact model compatibility is not independently tested.

Images and PDFs

Available

Images and PDFs can be sent in chat messages to every chat model. File size and model limits apply.

Hosted web search / browsing

Not exposed

No provider-hosted web search or web-fetch tool is exposed. Customer-owned search can be implemented as an authorized function tool.

Hosted code execution / computer use

Not exposed

No built-in execution environment or computer-control tool is provided. Code generation is not code execution.

Hosted file search / persistent assistants

Not exposed

No hosted file-search index, Assistants endpoint or persistent provider agent is exposed.

Prompt caching

Available

Gateway-reported cache reads and writes are included in billing. See this model’s cached input and cache-write rates; a cache hit is not guaranteed. Output is billed at the standard output rate.

Batch / fine-tuning

Not exposed

No public batch or fine-tuning endpoint is offered here.

Prompt caching

Cached input:: $0.10 per 1M tokens
Cache write:: No extra fee
Cached output:: Not offered · standard output rate applies

Eligible repeated prompt prefixes can use the cached-input rate when the gateway reports a cache hit. Cache hits are not guaranteed.

Caching guide and usage fields

TokenRex pricing

Input / 1M tokens

$2.00

Output / 1M tokens

$10.00

Cached input / 1M

$0.10

Cache write / 1M

No extra fee

Test in Playground Get API key

Usage-based billing. Repeated prompt prefixes are cached automatically and billed at the cached input rate.

Billing calculator

Total input tokensprompt_tokens, including cached tokensCached input tokenscached_tokens reported in usageOutput tokensIncludes thinking tokensNumber of requestsSame usage per request

Per request: $0.088
Total: $0.088
Without cache: $0.24

Cached tokens are billed at the cached-input rate; output is always billed at the standard rate. This is an estimate; your actual charge follows the usage the API reports.

API access

https://tokenrex.ai/api/public

POST /v1/chat/completions

Requests require a TokenRex API key and sufficient wallet balance.

Integration documentation

Streaming and customer-defined function tools are routed through the chat API; compatibility varies by model. Image/PDF input is accepted by the bridge, but model vision support needs provider confirmation. Hosted search and computer-use tools are not enabled; public Responses, Assistants, Batch and fine-tuning endpoints are not exposed.

Explore similar models

Gemini 2.5 Pro Gemini 3.1 Pro Gemini 3.5 Flash

Sources and verification

Provider specifications describe the upstream model, not a guarantee of every feature through TokenRex. Prices come from the enabled catalog; benchmark results belong to the exact tested model.

Gemini API · Gemini 4 Argon model specifications
Vendor-reported
Last checked 2026-10-14
Google DeepMind · Gemini 4 Argon model card
Vendor-reported
Last checked 2026-10-14
Google Blog · Introducing Gemini 4 Argon
Vendor-reported
Last checked 2026-10-14

Research notes (2)

Published provider limits; maximum-capacity requests have not been independently exercised through TokenRex. Standard requests allow up to 65,536 output tokens. Google's long-decode mode (output up to 1,048,576 tokens minus input) is enabled with background=true — the TokenRex chat endpoint accepts this parameter (verified live on 2026-10-14), but a full 1M-token output has not been exercised, so the default 65,536-token output limit is shown.
All benchmark scores are self-reported by Google in the official model card; TokenRex has not re-run them.