Gemini 3.8 Flash is reportedly undergoing internal testing at Google as rising AI costs increase demand for affordable models.
Business Insider reported seeing images showing “Gemini 3.8 Flash Preview” on Jetski, Google’s internal coding platform. An employee suggested it performed better than Gemini 3.7 Flash but cautioned that the assessment might be inaccurate.
Google declined to comment. Neither the model designation nor a public release date has been officially confirmed.
Google Accelerates Gemini Flash Releases and Cuts Pricing
Google introduced Gemini 3.6 Flash on July 21, followed by Gemini 3.7 Flash on August 13, just over three weeks later.
During Alphabet’s second-quarter earnings call, CEO Sundar Pichai said releases would arrive “almost at a monthly cadence” while Gemini 4 development continued.
Google describes Flash as a “workhorse” for coding and agents, where repeated model calls can generate substantial token costs.
Gemini 3.7 Flash’s introductory pricing runs through 2026 at $0.75 per million input tokens and $3.75 per million output tokens. Those rates halve Gemini 3.6 Flash’s launch prices of $1.50 and $7.50, respectively.
Gemini Flagship Delays Accompany AI Model Competition
On May 19, Google said Gemini 3.5 Pro was operating internally and would launch in June. By July 21, partner testing continued, with public availability deferred until readiness.
Google consequently continues competing on pricing and release speed while its flagship remains pending.
Artificial Analysis assigns Gemini 3.7 Flash an Intelligence Index score of 56 at high reasoning effort. GPT-5.6 Sol scores 61 at maximum effort, while Claude Fable 5 scores 62.
OpenAI’s Luna tier costs less, charging $0.20 per million input tokens and $1.20 per million output tokens. Across models, pricing ranges from cents to tens of dollars per million tokens, depending on capability.
Enterprise AI Spending Faces Greater Budget Scrutiny
Gartner forecasts global AI platform and model spending of $64.25 billion in 2026, up 63.4% from 2025.
Analyst Arunasree Cheparthi said enterprise budgets face “greater scrutiny,” with efficiency, cost control, and measurable outcomes receiving more attention.
Ramp data shows Claude Fable 5 represented 6% of businesses’ Anthropic token purchases during its first month. It generated 11.4% of Anthropic model spending despite launching atop Artificial Analysis’ intelligence ranking.
Anthropic charges $10 per million input tokens and $50 per million output tokens for Fable 5.
Ramp economist Ara Kharazian said the figures suggest a ceiling on business spending for raw capability. This favors affordable, fast models suited to specific tasks, the segment Google is targeting.
For now, information about Gemini 3.8 Flash remains based on Business Insider’s reporting. Google has not acknowledged the reported demonstration, leaving its eventual release uncertain as model names and launch schedules continue to change frequently.
