The newly introduced model addresses the sluggish 46 tokens per second throughput that hindered its predecessor, Gemini 4 Carbon, by reaching 127 tokens per second on OpenRouter. A predictive win-rate model for the Agent Duel leaderboard indicates that this substantial acceleration brings its projected Elo rating to 1,066. This performance places the system narrowly behind Opus 5.5, which recorded a score of 1,078. Overall, the analysis demonstrates that early voting behaviors heavily reward faster response times in head-to-head model evaluations.
Key points
Operational speed serves as a critical factor in competitive artificial intelligence rank…: Operational speed serves as a critical factor in competitive artificial intelligence rankings, especially among users who cast early votes.
The 90% confidence interval for the projected Elo difference between the two competing mo…: The 90% confidence interval for the projected Elo difference between the two competing models spans from -94 to +68.
Combining incremental benchmark gains with a triple-digit token rate allows Neon to overc…: Combining incremental benchmark gains with a triple-digit token rate allows Neon to overcome Carbon's adoption bottlenecks.
Sources checked by our analysis pipeline: 1
TLDR: we project Gemini 4 Neon’s performance based on what a faster Gemini 4 Carbon might score and find it equivalent to Opus 5.5
Gemini 4 Carbon is the best model on the Agent Duel leaderboard if you exclude early voting. Carbon’s main problem is that it is slooooow.
Today, Gemini 4 Neon was released, and looks to be an incremental benchmark improvement with a very large speed improvement. Earlier today, it was being served extremely fast via OpenRouter: 127 tokens per second. This is way faster than Carbon at 46 tokens per second.
This got me wondering, could Neon be not just a good model but the best model?
Modeling our leaderboard
To investigate, I modeled the win rate in our leaderboard. The best model has three components, model identity, speed and whether the user voted early. My initial modeling did poorly, so I added an interaction between speed and early voting. This ended up working well.
And surprisingly, speed seems to be pretty much as important as model identity (another blog on that soon).
Addressing widespread workforce anxiety surrounding automated systems, leadership offered voluntary buyouts to tens of thousands of staff to facilitate an internal reorganization around artificial intelligence. The strategy grants employees personal autonomy during sweeping workflow changes rather than imposing forced terminations. Ultimately, management expects all internal roles to transition from manual execution to directing and overseeing autonomous digital tools.
Capital secured from backers in Japan marks a major milestone for the artificial intelligence company as it completes its public listing. However, the final proceeds settled at the bottom of the targeted expectations outlined in earlier projections. The resulting total reflects the minimum threshold management had sought to achieve during this institutional offering phase.
The historic listing gives the San Francisco-based artificial intelligence firm a $1.5 trillion valuation as it prepares for its Nasdaq trading debut under the ticker symbol "ANT." This massive capital injection provides crucial funding to satisfy heavy infrastructure commitments without taking on additional debt. The milestone arrives despite mounting sector challenges, including fears of an industry bubble, public backlash, and looming congressional regulatory scrutiny.