Editorial illustration for Google's Gemini 3.8 Flash Priced at USD 0.58 Per Task
Google's Gemini 3.8 Flash: $0.58 Per Task
Google shipped Gemini 3.8 Flash on Thursday, three weeks after Gemini 3.7 Flash and just six weeks after the Flash line's last release. That's the third budget model in that span, and it comes in two flavors: a general-purpose reasoning and coding model, plus a specialized security build called 3.8 Flash Cyber. Pricing holds steady at $0.75 per million input tokens and $3.75 per million output tokens, matching what 3.7 Flash charged.
The pace raises an obvious question. Google still hasn't shipped Gemini 3.5 Pro or Gemini 4, the frontier models developers have been waiting on since before the Flash sprint started. Three budget releases in six weeks looks either like a company iterating fast on its cheapest tier, or like one filling the gap while its flagship work stalls.
New DeepMind head Koray Kavukcuoglu has pushed back on the idea that Google is only optimizing for price-performance now. On paper, 3.8 Flash backs that up: Google's benchmark numbers put it ahead of the previous Flash version by a wide margin on coding tasks, and close behind Anthropic's top model. Whether that's enough to quiet the frontier-model question is another matter.
Why this matters
Three Flash releases in six weeks tells us Google is optimizing for headlines about efficiency while its actual frontier models sit untouched. That's worth watching if you're building on Gemini: the "cheapest at its intelligence level" framing from Artificial Analysis sounds great until you notice the per-task cost jumped 40 percent over 3.7 Flash despite flat per-token pricing. That gap usually means the model is burning more tokens to reach an answer, which is a cost line that shows up on your bill even if the sticker price looks unchanged.
For founders running high-volume coding or cybersecurity workloads through 3.8 Flash Cyber, the lesson is to benchmark actual task cost, not headline pricing, before migrating off 3.7. And for anyone tracking Google's roadmap, the real question isn't whether Flash gets marginally better every three weeks, it's why the frontier tier has gone quiet while the budget line does all the talking. Cadence without a Pro or Ultra update alongside it starts to look like a pricing story more than a capability one.
Related Reading
Browse all Google news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic NVIDIA and Google Cloud let developers scale AI from prototype to production
Market Trends
Related Topic Google's FACTS benchmark shows 70% factuality ceiling across four tests
Research & Benchmarks
From Archives OpenAI's Astra Model Scores Perfect on Security Exploit Test
From Archives AI models can retrieve 65% of "inaccessible" facts by thinking longer
From Archives Google’s Gemini Agent Slashes Video Analysis Tokens by 88%
Research & Benchmarks
From Archives Google Pics Brings AI Image Generation to Google Workspace
Common Questions Answered
What is the pricing structure for Google's Gemini 3.8 Flash compared to the previous 3.7 Flash release?
Gemini 3.8 Flash maintains the same per-token pricing as 3.7 Flash at $0.75 per million input tokens and $3.75 per million output tokens. However, the actual per-task cost has increased by 40 percent despite the flat per-token pricing, indicating that the newer model requires more tokens to reach answers.
How does Gemini 3.8 Flash perform on the DeepSWE v1.1 benchmark for software engineering tasks?
According to Google's benchmarks, Gemini 3.8 Flash achieves 73.7 percent on the DeepSWE v1.1 benchmark for long-horizon software engineering tasks. This performance places it just below Claude Opus 5 at 74.0 percent and ahead of Claude Sonnet 5 (53.8%), GPT-5.6 Sol (72.7%), and its predecessor 3.7 Flash (65.3%).
What are the two variants of Gemini 3.8 Flash that Google released?
Google released Gemini 3.8 Flash in two flavors: a general-purpose reasoning and coding model for broad applications, and a specialized security build called Gemini 3.8 Flash Cyber designed for security-focused tasks. Both variants share the same pricing structure and release timeline.
Why is Google releasing multiple budget Flash models so frequently?
According to the article, Google appears to be optimizing for headlines about efficiency while its actual frontier models remain untouched. The rapid release cycle of three Flash models in six weeks suggests a strategic focus on budget-tier offerings rather than advancing its flagship frontier model capabilities.
Further Reading
- Gemini 3.8 Flash Models - Intelligence, Performance & Price - Artificial Analysis
- Gemini 3.8 Flash (low) API Provider Benchmarking & Pricing - Artificial Analysis
- Google Gemini API pricing August 2026 — official list & per ... - Kunavo
- Introducing Gemini 3.7 Flash - Google Blog
- Gemini API Pricing (September 2026): Model & Token Costs - BenchLM
