ModelsNov 19, 2026 · 18:188 sources

Google's latest AI model is faster, cheaper and smarter than its flagship predecessor from just weeks ago

From the-decoder.com

Image: the-decoder.com
AI;DR

Google DeepMind has released Gemini 4 Neon, its fourth model in seven weeks. The model is nearly 2x faster than its sibling Carbon, outperforms it on most benchmarks, and is available at a significantly lower price.

Read the sourceSign in to suggest

Sources

+ 6 more
  1. SUPPORT@kimmonismus“One of Googles small models surpassed the performance of their larger models within weeks and competes with Opus / Fable.”x.com
  2. SUPPORT@lentils80“Before release, it said Neon "builds on Gemini 4 Hydrogen" (RL'ing on it) and had "Flash-tier speed and cost." They cut both.”x.com
  3. SUPPORT@klieret“Gemini 4 Neon is claiming up to 39% on SWE-sweep, blowing all previous results out of the water. Very interesting that it strongly depends on the harness (the 12.8% with mini-swe-agent are not that big of a leap).”x.com
  4. SUPPORT@kimmonismus“for browser testing it sent out smaller Flash sub-agents to click through the demos. When those hit capacity errors, Neon asked me which model to try next.”x.com
  5. SUPPORT@pawelhuryn“I tested Gemini 4 Neon on real work. 2 repos, 105 replanted bugs. Find and fix what you can. It's the new #1. The results: - Gemini 4 Neon (xhigh): 63 - GPT-6.2 Astra (max): 62.7”x.com
  6. SUPPORT@art_zucker“$ANT just lost 10% on this new model”x.com

One ranked edition, every morning

The stories that matter today. Free, cancel anytime.

Telegram @aidr_today