Gemini 3.8 Flash scores 59 on AI Index at $0.58/task, Google's 4th Flash in 4 months
zacharynado · x · 2026-09-10
Artificial Analysis benchmarks Gemini 3.8 Flash, Google's fourth Flash model in under four months:
- Intelligence Index: 59 with high reasoning, up 3 points from Gemini 3.7 Flash, matching GPT-5.6 Sol (xhigh) and Grok 4.6 (medium); 57 at medium and 52 at low reasoning.
- Pricing: keeps 3.7 Flash's discounted rates through year-end ($0.75/$3.75 per M input/output tokens), sitting on the intelligence-vs-cost Pareto frontier at $0.58 per task — close to GPT-5.6 Terra (max, $0.53) but 40% pricier than its predecessor.
- Why cost rose: average output tokens per task up 30% to 48k, plus more turns on agentic evals.
- Gains driven by agentic work: biggest jump on 𝜏³-Banking (+12 pts to 45%), with improvements on Terminal-Bench v2.1 and GDPval-AA v2.
Related event: Gemini 3.8 Flash Underperforms in AA Intelligence Index(2 posts)→
More from Models
- Ruff creator Charlie Marsh: frontier models find infinite bugs in any software — charliermarsh · 2026-10-26
- Dev testing one skill across Opus 5.5 and Sol 6.1 finds instruction following 'all over the map' — lucasmeijer · 2026-10-26
- supermemory unveils learner-1, claiming continual learning for any model — julianweisser · 2026-10-26
- Anthropic says Opus 5.5 cuts cost per task by ~40% vs Opus 5 — taufiqintech · 2026-10-26
- AI search APIs ranked: top 12 within 7 points, yet prices differ nearly 7x — FinanceYF5 · 2026-10-26
- OpenAI Dots coming to EU/EEA and UK in a few weeks, slides confirm — testingcatalog · 2026-10-26