GoogleEco TierZero Data Retention

Gemini 3.7 Flash

Name: ARMES
Availability: InStock
Author: ARMES

Google's former workhorse model (released Aug 13, 2026; knowledge cutoff Mar 2026).

Run Gemini 3.7 Flash

Included on ECO plan

Context Window

1M in / 65K out

Input Token Price

$0.75

per 1,000,000 tokens

Output Token Price

$3.75

per 1,000,000 tokens

Speed Rating

Very

Very Fast (~340 t/s)

Verified Frontier Benchmarks

Empirical Evaluation Results

Evaluated across standard and verified harnesses

Artificial Analysis Intelligence Index

reasoning

DeepSWE v1.1

coding

65.3%

Terminal-Bench 2.1

coding

85.8%

FrontierCode 1.1 Main

coding

43.6%

AutomationBench

agentic

62.0%

GDPval-AA v2

agentic

1525 Elo

GDP.pdf

reasoning

34.0%

GDM-MRCR v2 128K

reasoning

97.0%

OSWorld-2.0

agentic

47.9%

WebDev Arena

reasoning

1588 Elo

GPQA Diamond

reasoning

94.5%

HLE

reasoning

47.9%

AA Coding Index

coding

76.1

AA Agentic Index

agentic

36.4

AA Intelligence Index

reasoning

39.1

Critpt

reasoning

14.3%

EnterpriseOps Gym

agentic

50.4%

AA GDPval 2.0

strategy

1406.6 Elo

Humanity's Last Exam (HLE)

reasoning

47.9%

Lcr (1.1)

reasoning

81.7%

Mmmu Pro

reasoning

85.5%

Omniscience

science

26.5

SciCode

science

57.2%

Tau3 Banking

finance

35.5%

Openrouter GPQA Diamond

reasoning

94.3%

Openrouter Tau2 Bench Airline

finance

80.6%

FrontierCode 1.1

coding

42.2%

DeepSWE / SWE-bench Pro

coding

65.5%

Context Arena MRCR V2 (8-needle)

agentic

96.0%

Surge Chartography (100-tasks)

reasoning

40.4%

Surge Gdp Pdf (100-tasks)

strategy

23.8%

TAU-Bench Banking

agentic

32.8%

Gdpval (2)

strategy

1406.6 Elo

TAU-3 Banking

agentic

32.8%

TAU-2 Airline

agentic

80.6%

Design Arena Frontend

coding

1214 Elo

Design Arena Fullstack

coding

1207 Elo

AnalystAgent

strategy

60.0%

Briefcase v1.1

strategy

1115.7 Elo

GDP PDF

strategy

23.6%

GDPval-AA v2.1

strategy

1433.1 Elo

Harvey LAB-AA

legal

90.7%

Mlcr

reasoning

15.0%

Terminal-Bench 4.0

coding

13.6%

Coding Agent Index (1.4)

agentic

59.6%

Livebench (2026-06-25)

reasoning

78.8

Vals EMB 2.0

finance

71.3%

Vals Vibe Code Bench

coding

70.4%

Vals Code Migration

coding

29.0%

Vals Legal Research

legal

34.6%

Apprenticebench API

reasoning

32.0%

Frontiermath Tiers 1 3 (v2)

math

71.6%

FrontierMath Tier 4

math

36.6%

SimpleQA

voice

69.2%

Chess Puzzles

reasoning

47.0%

Mystery Game Puzzles

reasoning

37.0%

Frontierswe (2)

coding

20.3%

Rsi Exam (0.1)

reasoning

0.3

Vals Index (2)

reasoning

59.3%

Vals Finance Agent

finance

59.0%

Vals Terminal Bench 2.1 (2)

coding

77.5%

AA-LCR v1.1

reasoning

81.7%

AA-Omniscience Index

reasoning

26.5

SimpleQA Verified

voice

69.2%

Terminal Bench (4.0)

coding

11.2%

Frontiercode Extended (1.1)

coding

54.7%

Furniture Assembly

finance

26.7%

Architectural Profile & Capabilities

Google's former workhorse model (released Aug 13, 2026; knowledge cutoff Mar 2026). Succeeded by Gemini 3.8 Flash. Algorithmic improvements over 3.6 Flash. Configurable thinking levels (low/medium/high). Computer Use in preview. Multimodal input (text/image/audio/video/PDF). Introductory pricing $0.75/$3.75 through Dec 31, 2026; standard $1.50/$7.50 thereafter. 90% cache discount ($0.075/M cached). Best for: agentic coding, iterative debugging, multi-step tool use, enterprise workflows

Architecture Type: Dense Transformer

Standard dense autoregressive transformer architecture with full-attention mechanisms.

Recommended Workloads & Primary Use Cases

Everyday reasoning

Drafting

Fast problem solving

ARMES Zero Data Retention

Calls to Gemini 3.7 Flash are routed through strict Zero Data Retention inference channels. Prompts and outputs are never stored, indexed, or monitored by Google.

View Privacy Architecture

More Models from Google

Compare other engines available in this laboratory.

View Roster (14)

Gemini 3.1 Flash TTSPro Tier

Gemini 3.1 Flash TTS is a neural speech synthesis model by Google, providing expressive voice rendering for real-time conversation.

128K ctx$1.00/M

Gemini 3.1 ProPro Tier

Abstract reasoning puzzles, competitive algorithms, agentic tool orchestration, scientific coding, large document analysis (10K+ tokens), technical/academic writing

1M ctx$2.00/M

Gemini 2.5 ProPro Tier

Legacy large-context analysis

1M ctx$1.25/M

Architectural Profile & Capabilities

Architecture Type: Dense Transformer

Standard dense autoregressive transformer architecture with full-attention mechanisms.