Gemini 3.8 Flash is Google's third budget model in six weeks while frontier models remain MIA
decoder ·
Google has released Gemini 3.8 Flash, its third budget model in six weeks, following Gemini 3.7 Flash by just three weeks. The new model comes in general-purpose and specialized cybersecurity versions, with Google reporting it surpasses its predecessor on coding benchmarks and nearly matches Claude Opus 5 on the DeepSWE v1.1 test. It is priced at $0.75 per million input tokens and $3.75 per million output tokens until January 2027, after which prices double, though it remains cheaper than rival frontier models. Google attributes the performance gains partly to extra reasoning steps and iterative tool use, which increase token consumption, so the company suggests lower reasoning levels or 3.7 Flash for efficiency-focused tasks. Independent testing by Artificial Analysis gives it an Intelligence Index score of 59, on par with GPT-5.6 Sol and Grok 4.6, but notes its cost per task rose about 40 percent versus 3.7 Flash. The cybersecurity version is only available through the Fairwind Program to vetted defenders and scores well on vulnerability detection and automated patching benchmarks while also showing strong resistance to prompt injection attacks.
decoder