57% vs 18.2% on Terminal-Bench 4.1.
57% vs 18.2% on Terminal-Bench 4.1. That gap, per Google, between Gemini 4 Hydrogen and Gemini 3.8 Flash is why I think the cheap workhorse model matters more to Google's AI business than another flagship.
- Google rolled Hydrogen out globally on November 12 across the Gemini API, AI Studio, Search, the Gemini app and eligible Antigravity tiers. It is built for fast repo-level coding and agentic workflows, priced at $1.50 per million input tokens and $7.50 per million output tokens.
- FourWeekMBA notes that is 2x Gemini 3.8 Flash's price. The Decoder reports Hydrogen outscores the bigger Argon just four weeks after Argon launched. Google is pricing it as a step up, and the benchmark gap gives it a case.
- Antigravity's Orchestrator switches between models like Carbon and Hydrogen mid-session. Google says that gets near-Carbon task success on Terminal-Bench 4.0 with 24% less quota and 19% faster completion. These are company numbers, so I want to see developer adoption before treating them as proven.
- $GOOG closed up 1.62% at $295.35 and traded at $296.25 after hours. I would not pin that whole move on one model launch.
My read: a sturdy mid-tier model is where agent budgets actually get spent, and Google keeps compounding there. Orchestrator usage is what I'm watching next. I could be wrong if the benchmarks don't hold up in real repos.
Sources:
Google — Gemini 4 Hydrogen announcement: https://blog.google/innovation-and-ai/technology/developers-tools/build-with-gemini-4-hydrogen/
AI Weekly — pricing & benchmarks: https://aiweekly.co/alerts/gemini-4-hydrogen-ships-at-150750-as-agent-daily-driver
GuruFocus — GOOG stock reaction: https://www.gurufocus.com/news/9195472/google-launches-gemini-4-hydrogen-ai-model-shares-rise-2-goog
Not financial advice. For informational and educational purposes only.