Physix Frontier · News Briefing Card (enbrief · Nov 20, 2026)

Google launches Gemini 4 Neon model as Antigravity coding tool draws complaints

KEY FACTS

  • Google released Neon, a new model in the Gemini 4 family, positioned between Carbon and Hydrogen.
  • Neon outputs at roughly 110 tokens per second, compared with about 57 tokens per second for Carbon.
  • Neon ranks first on the Artificial Analysis coding agent index.
  • Neon's API pricing is $3 per million input tokens and $15 per million output tokens.
  • Developers report that Antigravity suffers from slow responses, lost context on long tasks, and unpredictable quota consumption.

KEY DATA

about 110 tokens/sNeon output speed
about 57 tokens/sCarbon output speed
$3 per million tokensNeon API input price
$15 per million tokensNeon API output price

PHYSIX OBSERVATION

No matter how impressive the model's benchmark scores are, a lagging toolchain drags it down. Neon's speed and coding ability are already close to flagship level, but Antigravity's stability and quota mechanism make it hard for developers to cash in on those capabilities. Google's weak spot is not the model but engineering delivery, which is exactly what users actually pay for.

Source: enbrief original report ↗