Breaking

Google DeepMind launches Gemini 3.8 Flash for coding and agents at 3.7 Flash prices

September 2, 2026 at 12:49 EDT

Foundation Models
Software Dev & Coding
AI Agents

Google DeepMind has released Gemini 3.8 Flash, a generally available workhorse model aimed at long-horizon software engineering and autonomous agents, at the same introductory price as Gemini 3.7 Flash. The company also introduced Gemini 3.8 Flash Cyber, a restricted cybersecurity sibling for trusted defenders.

Sep 2, 2026 · Google DeepMind

Gemini 3.8 Flash lands

Google's third Flash model in six weeks targets long-horizon coding and autonomous agents at the same introductory price as Gemini 3.7 Flash.

Glean long-running tasks vs 3.7 Flash

1×

3.7 Flash

3×+

3.8 Flash

Glean reported more than three times as many completed long-running document tasks versus Gemini 3.7 Flash.

3rd

Flash release in 6 weeks

$0.75

per 1M input tokens (intro)

54.9%

HLE-Verified

input context tokens

Input price per 1M tokens

$0.75

Intro through 2026

$1.50

Standard from 2027

Standard input pricing is 2× the introductory rate once the discount expires on January 1, 2027.

What teams like

Same intro price as 3.7 Flash, stronger coding and agent scores, and a 1M-token context with computer use and code execution.

The catch

It uses more tokens on hard jobs—about 40% higher cost per task in one independent index—plus possible timeouts and a mixed knowledge cutoff.

Where it runs

Generally available today for developers and paying Gemini subscribers.

Gemini API · gemini-3.8-flash AI Studio · Antigravity default Gemini app · AI Mode · Sheets Flash Cyber · Fairwind only

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.