Google DeepMind Releases Gemini 4 Neon: 1M Context, #1 on AA Coding Agent Index
Google DeepMind released Gemini 4 Neon, a reasoning model positioned between Gemini 4 Carbon and Gemini 4 Hydrogen and aimed at coding agents and knowledge work. The release claims a 1M-token context window and the #1 spot on the AA Coding Agent Index. Separately, Google introduced the multimodal decision model Gemini Decision 1, which leads on five benchmarks and trails on two. Google also published a research paper, "Harnesses That Improve Themselves," showing that an AI agent that gradually modifies its own runtime harness and validates each change can substantially raise task success rates.
How this story unfolded
Key points7 items Model summary · ordered by when the evidence was published
- Point 01 11/19, 14:52
Google DeepMind released the reasoning model Gemini 4 Neon, positioned between Gemini 4 Carbon and Gemini 4 Hydrogen.
Based on 1 - Point 02 11/19, 14:52
Gemini 4 Neon is aimed at coding agents and knowledge work tasks.
Based on 1 - Point 03 11/19, 14:52
Gemini 4 Neon has a 1M-token context window and is ranked #1 on the AA Coding Agent Index.
Based on 1 - Point 04 11/19, 17:08
Google introduced Gemini Decision 1, a multimodal decision model, which leads on five benchmarks and trails on two.
Based on 1 - Point 05 11/19, 17:08
Google said the decision model shows the field is not a bubble but a category, referencing Jev's $7.5 billion valuation.
Based on 1 - Point 06 11/20, 08:19
Google published a research paper titled "Harnesses That Improve Themselves."
Based on 1 - Point 07 11/20, 08:19
The paper shows letting an AI agent gradually modify its own runtime harness and validating each change can substantially raise task success rates.
Based on 1
ReportsBy publish time · newer further down
- November 19, 2026 · Thu
- Start
Google DeepMind Releases Gemini 4 Neon: 1M Context, #1 on AA Coding Agent Index
Original summary Google DeepMind just released Gemini 4 Neon . Gemini 4 Neon is a reasoning model that sits between Gemini 4 Carbon and Gemini 4 Hydrogen. It targets coding agents, knowledge work a
MarkTechPost Original ↗ - +2h
Google enters Jev's core arena: Gemini Decision 1 leads on five benchmarks, trails on two
AI overview Jev's $7.5 billion valuation was once questioned as a bubble. Now Google is telling the market with a multimodal decision model: this isn't a bubble, it's a category. Google has of
雷锋网 Original ↗ - November 20, 2026 · Fri
- +17h
One Code Change Lets Google AI Agent Boost Task Success Rate 8x After Harness Optimization
AI overview A new Google research paper, "Harnesses That Improve Themselves," shows that letting an AI agent gradually modify its own runtime harness and validating each change can significant
IT之家 Original ↗ - Still developing · later reports will keep extending this axis