Gemini 3.8 Flash: three points up, forty percent more per task

Google has released Gemini 3.8 Flash, its fourth Flash model in under four months - it scores 59 on the Artificial Analysis Intelligence Index and reaches the Intelligence vs. Cost per Task Pareto frontier @GoogleDeepMind released Gemini 3.8 Flash today. With high reasoning, it scores 59 on the Artificial Analysis Intelligence Index, up 3 points from Gemini 3.7 Flash and on par with sub-maximum reasoning efforts of GPT-5.6 Sol (xhigh, 59) and Grok 4.6 (medium, 59) Matching Gemini 3.7 Flash’s discounted pricing until the end of the year ($0.75/$3.75 per million input/output tokens), Gemini 3.8 Flash sits on the Intelligence vs. Cost per Task Pareto frontier at $0.58 per task. This is comparable to GPT-5.6 Terra (max, $0.53), but ~40% higher than its predecessor, driven by a 30% increase in average output tokens per task to 48k and increased turns on agentic evaluations Key benchmarking results across Gemini 3.8 Flash’s three reasoning levels: ➤ 3 point Intelligence Index improvement: Gemini 3.8 Flash (high) scores 59 on the Artificial Analysis Intelligence Index, up 3 points from Gemini 3.7 Flash (high, 56). With medium reasoning it scores 57, matching GPT-5.6 Terra (max, 57) and Muse Spark 1.2 (xhigh, 57). With low reasoning it scores 52, matching Gemini 3.6 Flash (high, 52), at 30% lower Cost per Task and roughly a third of the Time per Task ➤ Agentic capability improvements: Gemini 3.8 Flash’s 3 point improvement on the Artificial Analysis Intelligence Index is primarily driven by stronger performance on agentic evaluations such as 𝜏³-Banking (tool use), Terminal-Bench v2.1 (coding) and GDPval-AA v2 (real-world tasks). The largest improvement is on 𝜏³-Banking, where it gains 12 points over Gemini 3.7 Flash to score 45% ➤ Pareto frontier on Intelligence vs. Cost per Task: Gemini 3.8 Flash (high) costs $0.58 per Intelligence Index task, making it the cheapest model at its level of intelligence. This is up ~40% from Gemini 3.7 Flash ($0.40) despite unchanged per-token pricing, driven by a 30% increase in output tokens per task and more turns on agentic evaluations. Cost per Task falls to $0.41 with medium reasoning and $0.24 with low reasoning ➤ Output speeds remain fast, but Time per Task increases: On high reasoning, Gemini 3.8 Flash averages ~300 output tokens per second and a Time per Task of 2.5 minutes, slightly faster than GPT-5.6 Luna (max, 2.6 minutes) and GPT-5.6 Terra (max, 3.3). Compared to Gemini 3.7 Flash, higher token usage increases Time per Task from 2.2 minutes to 2.5 minutes, and puts it behind Claude Fable 5.1 (medium, 2.1 minutes). On low reasoning, Time per Task falls to 0.8 minutes, placing Gemini 3.8 Flash on the Intelligence vs. Time per Task Pareto frontier Key model details: ➤ Context Window: 1M tokens, unchanged from Gemini 3.7 Flash ➤ Multimodality: Text, image, video, and speech input, with text output ➤ Pricing: $0.75/$3.75 per 1M input/output tokens through the end of the year, matching Gemini 3.7 Flash’s current discounted pricing. $1.50/$7.50 per 1M input/output tokens at standard pricing. Cached input tokens retain the same 90% discount

Gemini 3.8 Flash (high) costs $0.58 per Intelligence Index task, the cheapest we’ve measured at this level of intelligence. This is up ~40% from Gemini 3.7 Flash despite unchanged per-token pricing, driven by a 30% increase in output tokens per task and more turns on agentic evaluations. Cost per Task falls to $0.41 with medium reasoning and $0.24 with low reasoning. Gemini 3.8 Flash is priced at $0.75/$3.75 per 1M input/output tokens until the end of the year, matching Gemini 3.7 Flash’s discounted pricing, with cached input tokens retaining the same 90% discount versus standard input

On high reasoning, Gemini 3.8 Flash averages ~300 output tokens per second and a Time per Task of 2.5 minutes, just ahead of GPT-5.6 Luna (max, 2.6 minutes) and GPT-5.6 Terra (max, 3.3). Compared to Gemini 3.7 Flash, higher token usage increases Time per Task from 2.2 minutes to 2.5 minutes, and puts it behind Claude Fable 5.1 (medium, 2.1 minutes). On low reasoning, Time per Task falls to 0.8 minutes, placing Gemini 3.8 Flash on the Intelligence vs. Time per Task Pareto frontier

With high reasoning, Gemini 3.8 Flash averages 48k output tokens per task, a 30% increase compared to 3.7 Flash. On low reasoning, Gemini 3.8 Flash outputs an average of 14k tokens per task, just under GP-5.6 Sol with max reasoning (17k)