Google says its new Gemini 3.8 Flash model ‘works harder’ but might cost more

Google has launched Gemini 3.8 Flash, which performs more reasoning steps on complex tasks and calls tools iteratively compared to its predecessor. Although it retains the introductory pricing of $0.75 per million input tokens and $3.75 per million output tokens, Google warns it might use more tokens to maximize performance, raising costs. Artificial Analysis noted the model is extremely cost-effective for its intelligence level, though total task costs are up about 40% due to increased output tokens and agentic evaluation turns. Industry experts praised its coding capabilities and high execution speed. Google reports significant improvements for software engineering and autonomous agents, outperforming competing frontier models on major benchmarks including DeepSWE v1.1, Vals Finance Agent V2, and Harvey’s Legal Agent. The model also incorporates safeguards against misuse across sensitive scientific and technical domains.