One post tagged with "Gemini"

Google released Gemini 3.8 Flash on September 2, 2026, calling it its most intelligent Flash model and making it generally available for production use. The launch also includes Gemini 3.8 Flash Cyber, a trusted-access model for cybersecurity defenders.

Gemini 3.8 Flash is not trying to win by being the biggest model. Its pitch is a combination of a 1M-token context window, multimodal input, built-in Google grounding, configurable thinking, fast serving, and an introductory API price of $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026.

That price changes the comparison with Claude. Claude Fable 5.1 is the closest match in long-horizon ambition, but it costs $10/$50 per million tokens. Claude Opus 5 costs $5/$25, while Claude Sonnet 5 costs $2/$10 in Anthropic's current model overview. Gemini 3.8 Flash is therefore much cheaper on token price, but the comparison cannot stop at price: Google explicitly says the model may use more reasoning tokens and more tool calls on difficult work.

The early community signal is encouraging but not conclusive. Users report that 3.8 Flash feels less rushed and more thorough than 3.7 Flash, while others report higher latency, heavier quota consumption, inconsistent rollout, and little quality improvement on small evaluation sets. The right question is not “Is Gemini 3.8 Flash smarter than Claude?” It is: which model gives your workflow the best completed result per dollar, minute, and human review cycle?