The LogiciansStrict ZDR Verified

DeepSeek

Name: ARMES
Availability: InStock
Author: ARMES

Versatile reasoning at the lowest cost.

Run DeepSeek in ARMES

Laboratory Overview

DeepSeek is an open research powerhouse known for pioneering breakthrough Mixture-of-Experts architectures like Multi-Head Latent Attention (MLA) and DualPipe MoE. Their V4 Pro and R1 models offer frontier math and coding at unprecedented cost efficiency.

Core Research Focus:Pure logicmathefficiencyversatile general-purpose intelligence at extremely low cost.

Privacy & Zero Data Retention

Strict Zero Data Retention (ZDR): Requests are processed ephemerally in RAM. No user data, prompts, or completions are stored, indexed, or used for model training.

No user telemetry sent to DeepSeek

DeepSeek Foundation Models (7)

Available under ARMES unified subscription without separate API keys or billing agreements.

Tier:

Mixture of Experts (MoE)

MoEEco

V4 Pro 0813

Terminal-Bench 2.1: 87.9%, DeepSWE: 62.7%, CyberGym: 83.3%, AutomationBench: 31.8%, Toolathlon-Verified: 74.1%, HLE w/ tools: 60.0%, GPQA Diamond: 90.1%, Artificial Analysis Intelligence Index: 53.2, AA Coding Index: 68.8, AA Agentic Index: 49.6 | The official GA release of V4 Pro, superseding the April preview (released Aug 13, 2026).

AA Intelligence Index:36.0

1M ctx

$1.32 / $3.96 /MTok

Intelligence Dossier Try in Chat

Mixture of Experts (MoE)

MoEEco

V4 Pro

V4 Pro is a high-performance foundation model engineered by DeepSeek, accessible with Zero Data Retention on ARMES.

AA Intelligence Index:30.4

128K ctx

$1.74 / $3.48 /MTok

Intelligence Dossier Try in Chat

Sparse Mixture-of-Experts (MoE) Transformer — Causal Encoder-Decoder (CED), multimodal

MoEFree

V4.1 Flash

DeepSeek-V4.1-Flash is DeepSeek's first Causal Encoder-Decoder (CED) frontier model: a 552B-parameter multimodal sparse MoE that activates only 8B parameters during prefill and 16B during decode, with a 1M-token context window, native image understanding, and aggressive FP4/compressed KV caching. It tops agentic terminal and software-engineering boards (90.6 on Terminal-Bench 2.1, 74.2 on DeepSWE v1.1 at max effort) at roughly $0.15/$0.60 per 1M tokens, and supersedes V4-Pro in production traffic.

Terminal-Bench 2.1 (max effort):90.6%

1.05M ctx

$0.30 / $1.20 /MTok

Intelligence Dossier Try in Chat

Mixture of Experts (MoE)

MoEFree

V4 Flash-0731

Terminal-Bench 2.1: 82.7%, CyberGym: 76.7%, Toolathlon-Verified: 70.3%, DSBench-FullStack: 68.7%, DSBench-Hard: 59.6%, DeepSWE: 54.4%, NL2Repo: 54.2%, Agents' Last Exam: 25.2%, AutomationBench Public: 25.1%, Artificial Analysis Intelligence Index: 50 | The official GA release of V4 Flash, superseding the April preview (released Jul 31, 2026).

AA Intelligence Index:34.3

1M ctx

$0.14 / $0.28 /MTok

Intelligence Dossier Try in Chat

Mixture of Experts (MoE)

MoEFree

V4 Flash

Strong on long-document and routine coding | Cost-effective high-volume tasks, long-document processing, routine coding, local deployment on consumer hardware.

AA Intelligence Index:24.2

1M ctx

$0.14 / $0.28 /MTok

Intelligence Dossier Try in Chat

Mixture of Experts (MoE)

MoEEco

— | Deep reasoning, chain-of-thought tasks

Matharena Apex (2025):1.0%

128K ctx

$0.70 / $2.50 /MTok

Intelligence Dossier Try in Chat

Mixture of Experts (MoE)

MoEFree

V3.2

GPQA Diamond: 82.4%, SWE-bench: 73.1%, AIME: 93.1% | General chat, simple Q&A, translation, notes retrieval, simple scripts, moderate analysis, standard tasks.

AA Intelligence Index:21.5

128K ctx

$0.56 / $1.68 /MTok

Intelligence Dossier Try in Chat