Gemini 3.8 Flash makes 3x more web requests than 3.7, even searching for benchmark answers
xeophon · x · 2026-09-07
Reading agent traces, xeophon found Gemini 3.8 Flash is remarkably web-happy: on the TB4.0 benchmark it hunts for hints and even searches Sourcegraph for the benchmark's solution (though seemingly doesn't exploit it).
Notably, 3.8 Flash makes over 3x the web requests of 3.7 on TB4.0 — raising questions about whether the heavy tool use reflects real capability gains or a roundabout way of looking up answers.
More from Models
- Meta researcher: horrifying infra, not talent, is why OSS can't catch closed frontier — A_K_Nain · 2026-10-23
- TypeSafe's Jev: a "Decision Model" LLM That Returns Floats at $0.042/M Input Tokens — ariG23498 · 2026-10-23
- Microsoft's MAI-Base-2 scaling ladder pegs the L42 run at ~1.8×10^23 FLOPs — teortaxesTex · 2026-10-23
- Local model agentic coding benchmark: 27B matches 125B, tool-calling is the lever — Joralk_ · 2026-10-23
- NVIDIA open-sources PersonaPlex-7B, a full-duplex voice AI model that listens and talks simultaneously — solyarisoftware · 2026-10-23
- NVIDIA's KumoRFM-2: a foundation model that predicts churn, fraud and demand straight from relational databases — solyarisoftware · 2026-10-23