What are the strengths of this model?

Long-horizon coding near frontier level: DeepSWE v1.1 73.7% vs Claude Opus 5's 74.0% Terminal-Bench 2.1 jumps to 90.8% from 3.7 Flash's 81.6% 1M context, 65,536-token output, multimodal input, and three configurable thinking levels Intro price of $0.75/$3.75 through 2026 with default rollout across Gemini surfaces

What are the weaknesses of this model?

Intro pricing doubles January 1, 2027 and higher per-task token use can raise real costs ~40% Still trails top models on the hardest agentic tests (Terminal-Bench 4.0-class) No audio/image generation or Live API; computer use remains preview Quiet model-page launch and a ~30-minute listing removal suggest operational roughness

What are the best use cases?

Long-horizon software engineering and autonomous coding agents Document-heavy enterprise workflows over a 1M-token context High-volume agentic API workloads with multimodal input Cost-neutral migration off Gemini 3.7 Flash before 2027 pricing doubles

Google DeepMindProprietary

Gemini 3.8 Flash

Name: Gemini 3.8 Flash
Author: Google DeepMind

Compare this model

Google DeepMind's most intelligent Flash model, generally available September 2, 2026 and the third Flash-tier release in 43 days. Gemini 3.8 Flash improves on every published benchmark versus Gemini 3.7 Flash, posting 73.7% on DeepSWE v1.1 (Claude Opus 5: 74.0%, GPT-5.6 Sol: 72.7%) and 90.8% on Terminal-Bench 2.1 (3.7 Flash: 81.6%), and is engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows. It offers a 1,048,576-token context window, 65,536-token output, and multimodal input (text, image, video, audio, PDF) with text output. API pricing is an introductory $0.75/$3.75 per million tokens, doubling on January 1, 2027.

Share:X Hatena LINE

Parameters

Undisclosed

Context Window

License

Proprietary

Release Date

2026-09-02

API Pricing

Input Price (per 1M tokens)

$0.75

Output Price (per 1M tokens)

$3.75

Billing Mode: standard

Strengths

•Long-horizon coding near frontier level: DeepSWE v1.1 73.7% vs Claude Opus 5's 74.0%
•Terminal-Bench 2.1 jumps to 90.8% from 3.7 Flash's 81.6%
•1M context, 65,536-token output, multimodal input, and three configurable thinking levels
•Intro price of $0.75/$3.75 through 2026 with default rollout across Gemini surfaces

Weaknesses

•Intro pricing doubles January 1, 2027 and higher per-task token use can raise real costs ~40%
•Still trails top models on the hardest agentic tests (Terminal-Bench 4.0-class)
•No audio/image generation or Live API; computer use remains preview
•Quiet model-page launch and a ~30-minute listing removal suggest operational roughness

Use Cases

•Long-horizon software engineering and autonomous coding agents
•Document-heavy enterprise workflows over a 1M-token context
•High-volume agentic API workloads with multimodal input
•Cost-neutral migration off Gemini 3.7 Flash before 2027 pricing doubles

Deep Analysis

Release

September 2, 2026 (GA)

Model ID gemini-3.8-flash; third Flash launch in 43 days

Context Window

1,048,576 tokens

65,536-token output ceiling; text/image/video/audio/PDF input, text output

Intro Price

$0.75 / $3.75 per 1M

Through Dec 31, 2026; doubles to $1.50/$7.50 on Jan 1, 2027; cache inputs ~90% off

DeepSWE v1.1

73.7%

vs Claude Opus 5's 74.0% and GPT-5.6 Sol's 72.7%

Terminal-Bench 2.1

90.8%

Up from 81.6% on Gemini 3.7 Flash

Strengths

・Near-frontier long-horizon coding at a budget price: DeepSWE v1.1 73.7% essentially matches Claude Opus 5 (74.0%).
・Third Flash release in 43 days means the iteration loop itself is the product — 3.8 Flash inherits a mature, migration-tested line.
・1M-token context with true multimodal input (text, image, video, audio, PDF) and three configurable thinking levels.
・Intro price of $0.75/$3.75 holds through 2026 and rolls out as the default in the Gemini app, AI Mode, and Sheets for Pro/Ultra.

Weaknesses

・Intro pricing doubles on January 1, 2027, and the model consumes more reasoning tokens per task — real job costs can run ~40% higher than 3.7 Flash despite equal per-token prices.
・Still trails the top models on the hardest agentic tests (Terminal-Bench 4.0, OSWorld 2.0, GDPVal-AA v2).
・No audio or image generation, no Live API; computer use remains in preview.
・Google shipped it quietly via a model page rather than an announcement, and briefly pulled the listing for ~30 minutes before re-confirming — operational polish still lags the model quality.

Competitor Comparison

Model	Arena	SWE	GPQA	Price
Gemini 3.8 Flash	73.7% (DeepSWE v1.1)	90.8% (Terminal-Bench 2.1)	N/A	$0.75/$3.75
Claude Opus 5	74.0% (DeepSWE v1.1)	N/A	N/A	$15/$75
GPT-5.6 Sol	72.7% (DeepSWE v1.1)	N/A	N/A	$5 input
Gemini 3.7 Flash	65.3% (DeepSWE v1.1)	81.6% (Terminal-Bench 2.1)	N/A	$0.75/$3.75

Overview

Gemini 3.8 Flash, generally available September 2, 2026, is Google DeepMind's third Flash-tier release in 43 days and its most intelligent Flash model to date, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows. At an intro price of $0.75/$3.75 per million tokens it posts 73.7% on DeepSWE v1.1 — within a point of Claude Opus 5 — making frontier-adjacent agentic coding available at a fraction of the cost.

Benchmarks & Performance

Google reports Gemini 3.8 Flash at 73.7% on DeepSWE v1.1 (vs Claude Opus 5's 74.0% and GPT-5.6 Sol's 72.7%) and 90.8% on Terminal-Bench 2.1, up from 81.6% on 3.7 Flash. It improves on every published row versus 3.7 Flash, with the largest gains on long-horizon and agentic benchmarks. It still trails the top models on the hardest tests — Terminal-Bench 4.0, OSWorld-2.0, and GDPVal-AA v2 Elo.

Detailed Comparison

Against Claude Opus 5, 3.8 Flash nearly matches DeepSWE v1.1 at roughly 5% of the price, though Opus 5 still wins the hardest agentic rows by wide margins. Versus GPT-5.6 Sol it leads on DeepSWE v1.1 (73.7% vs 72.7%). Against its direct sibling Gemini 3.7 Flash, it improves on every published benchmark at identical per-token pricing — but uses more tokens per task, so real cost per job rises. The September 2 launch was overshadowed by GPT-6 Astra's September 3 unveiling.

Community Feedback

Developer reaction split between enthusiasm for the price-performance and irritation at a silent launch — Google updated a model page rather than issuing a release post, and briefly pulled the listing for about 30 minutes before re-confirming it. Analysts read the rapid Flash cadence as a response to investor pressure while Gemini 3.5 Pro remains delayed. A companion Gemini 3.8 Flash Cyber variant is gated behind Google's Fairwind Program for vetted security teams.

Use Cases

The economic case for 3.8 Flash is high-volume agentic coding and enterprise document workflows where a million-token context and multimodal input do most of the work: migration off 3.7 Flash is cost-neutral at per-token level until January 2027, so teams should measure actual task cost now before the intro price doubles. For hardest-tier agentic tests (Terminal-Bench 4.0-class), budget for a frontier fallback model. The Cyber sibling, restricted to vetted defenders, is a separate access decision for security teams.

Latest News

Google made Gemini 3.8 Flash generally available on September 2, 2026, its third Flash-tier release in 43 days (3.6 Flash on July 21, 3.7 Flash on August 13). Introductory pricing of $0.75/$3.75 per million tokens runs through December 31, 2026 and doubles to $1.50/$7.50 on January 1, 2027. The companion Gemini 3.8 Flash Cyber variant is restricted to trusted defenders under the Fairwind Program, which Google says now spans 650+ partners.

Sources

Analysis generated: 2026-09-04