Swapping to Antigravity CLI Lifted Gemini 4 Argon to Near Opus 5.5 in Real-Work Bug Benchmark
PawelHuryn · x · 2026-10-14
PawelHuryn found that running Gemini 4 Argon with Antigravity CLI instead of Gemini CLI dramatically changed its score: from "good" to frontier-level, landing very close to Opus 5.5 (max), confirmed across three runs.
On his hidden benchmark of 105 real bugs: GPT-6 Astra scored 45, Fable 5.1 43, Opus 5.5 41.7, and Gemini 4 Argon 33 — far above its earlier Gemini CLI result. The lower cost came from fewer tokens and more cache hits. He notes the model wasn't available on agy CLI and plans to retest.
More from coding & agent
- ServiceNow and McGill's AgentHorizon benchmark shows LLM judges miss subtle agent failures — rohanpaul_ai · 2026-10-14
- Sazabi launches 1.0: an AI agent that reads your logs and alerts with root cause and fix — rohanpaul_ai · 2026-10-14
- OpenCrewBot: open-source Grok Bot alternative runs your own team of AI bots — Arindam_1729 · 2026-10-14
- Jev via Jinja-templated Skills cuts Claude context cost 70% and runs 4-5x faster — oleg_ivye · 2026-10-14
- Dev behind $1.27M AI DM operation details prompt drift and silent API degradation — alanshore222 · 2026-10-14
- Open-source MCP verifies Australian ABNs against the live Business Register — Impossible_Bit_2676 · 2026-10-14