Google Challenges OpenAI, Anthropic with Gemini 3.8 Flash at Lower Cost

Google is gaining ground in the AI race. After lagging behind OpenAI and Anthropic, the company has launched Gemini 3.8 Flash, asserting that it is its most advanced AI model in terms of programming performance. Google claims that Gemini 3.8 Flash can rival advanced models like Claude Opus 5 and GPT-5.6 Sol, but at a significantly lower cost.

Google CEO Sundar Pichai announced the launch on X, noting that this model represented the company's "third Flash series launch" in just six weeks. Flash models are smaller; they are typically trained on hundreds of billions of parameters, rather than the trillions used by larger models. Furthermore, they are more cost-effective to run.

Pichai stated that, according to performance tests, Gemini 3.8 Flash "outperforms most larger, state-of-the-art models in autonomously solving complex engineering problems from start to finish, and at a fraction of the cost." He added that the model offers significant improvements over Gemini 3.7 Flash in tasks such as software engineering and multi-step reasoning.

Alongside the standard Flash model, Google also launched Gemini 3.8 Flash Cyber, describing it as its most capable cybersecurity model. Gemini 3.8 Flash is more cost-effective to run

The company is offering an introductory price of 75 cents (approximately 70 rupees) per million input tokens and $3.75 (approximately 350 rupees) per million output tokens through December 31, 2026. However, Google warns that the model may consume more tokens for highly complex tasks. Starting December 31, rates will rise to $1.50 (approximately 140 rupees) per million input tokens and $7.50 (approximately 700 rupees) per million output tokens.

For comparison, Anthropic’s Claude Opus 5 would cost $5 (approximately 470 rupees) per million input tokens and $25 (approximately 2,350 rupees) per million output tokens. Meanwhile, OpenAI’s GPT-5.6 Sol would cost $4 (approximately 375 rupees) per million input tokens and $20 (approximately 1,890 rupees) per million output tokens.

Cost has become an increasingly important factor in the AI race as companies review their spending on this technology. Lower costs could give Google an edge in the enterprise market, where programming remains a key use case. However, Chinese open-weight models, such as Kimi K3, have also recently gained ground by being even more cost-effective to run.

Gemini 3.8 Flash achieves significant breakthroughs

In the HLE-Verified test, Google reported that Gemini 3.8 Flash achieved a score of 54.9%, which the company says demonstrates its ability to handle multi-step reasoning across areas such as STEM disciplines, the humanities, and professional fields. Its knowledge cutoff date is March 2026 for some domains, while for others, users may find information limited to January 2025.

Google explained that these improvements stem from a design decision allowing 3.8 Flash to "work harder" by executing additional reasoning steps and iteratively invoking tools for complex tasks. The company notes that this could raise total usage costs, even with token rates remaining unchanged; it also indicates that developers seeking to reduce token spending can opt for lower-intensity configurations or continue using Gemini 3.7 Flash. Artificial Analysis stated that Gemini 3.8 Flash is "the most economical model we have measured at this intelligence level," though it added that total usage increased by approximately 40% compared to version 3.7 Flash, due to a higher volume of output tokens and a greater number of iterations in autonomous agent evaluations.

Gemini 3.8 Flash is now available in the Gemini app for subscribers of Google AI Pro and Ultra plans. It is also accessible to developers via Google Antigravity, AI Studio, and the Gemini API.

Google Launches Gemini 3.8 Flash Cyber

Google also announced Gemini 3.8 Flash Cyber, which replaces the 3.5 Flash Cyber version for trusted defenders under the new Fairwind program. The company stated that the model offers "state-of-the-art performance in autonomous vulnerability discovery" and focuses on detection and patching rather than offensive capabilities.

Google noted that the Chrome security team found Gemini 3.8 Flash Cyber generated 2.6 times more correct patches for Chrome vulnerabilities than the best larger commercial models. Google's cloud vulnerability research team reported using the model to discover a critical vulnerability in under two hours—a task that typically takes months.

This move brings Google closer to its competitors in the field of AI applications for cybersecurity. In recent weeks, Anthropic and OpenAI have made significant strides with cybersecurity-focused models. OpenAI recently claimed that its upcoming model, Astra, had surpassed critical cybersecurity benchmarks.

The launch of Gemini 3.8 Flash and 3.8 Flash Cyber comes as Google seeks to strengthen its position against OpenAI and Anthropic in the realm of AI coding tools. The tech giant had lagged behind its rivals and reportedly even delayed the release of its flagship model, Gemini 3.5 Pro. Meanwhile, the company has undergone major leadership changes in its AI division, including the departure of Demis Hassabis (of DeepMind) from his role as CEO. Google is reportedly currently working on a new flagship model: Gemini 4.

ABOUT THE AUTHOR

Kahekashan

[email protected]

Kahekashan is a passionate technophile with a keen eye for cutting-edge gadgets, emerging technologies, and everything in the digital realm. Raised in a Defence family with strong values and a background in literature, she has consistently pursued excellence in every endeavour. Her last full-time assignment involved content writing with the Indian School of Business.

Read More