Vibeleaderboard
← All Intel
Intel / article

Gemini 4 Neon Intelligence Gain

Source
artificialanalysis.ai
Author
Artificial Analysis articles
Date

Context

Artificial Analysis, an independent group that benchmarks AI models, found that Gemini 4 Neon scores only slightly higher than Google's Gemini 4 Carbon, the model Google positions for its hardest work, and that Neon's lower cost and time per task come from lower prices and faster output rather than from using fewer tokens. In the group's write-up, Neon at its xhigh setting scores 61 on the Artificial Analysis Intelligence Index, one point above Carbon at high and third overall, behind GPT-6.2 Astra and Claude Fable 5.2. At high, Neon matches Carbon's score at less than half the cost per task, a comparison made at Carbon's standard price.

The cost comparison is worth reading closely. Running the index's tasks costs $6.44 per task with Neon at xhigh against $7.76 for Carbon, about 17% less, and takes about 21 minutes per task instead of about 36. But Neon uses about 15% more output tokens per task than Carbon. Artificial Analysis attributes the whole cost gap to Neon's prices, $3 per million input tokens and $15 per million output tokens against Carbon's standard $4 and $20, and the time gap to its output speed of about 110 tokens per second against about 57. Those Carbon figures use its standard price. Carbon is currently discounted to $2 and $10 until November 30, and at that rate each Carbon token costs a third less than a Neon token.

For coding, the group measures each model inside the tool that runs it. Running in Google's Antigravity CLI, Neon at xhigh scores 59 on the group's Coding Agent Index, against 55 for Carbon and 58 each for Claude Code running Anthropic's Claude Fable 5.2 or Claude Opus 5.5. Its average cost per task in Antigravity is $6.86, against $12.61 for Opus 5.5 and $15.47 for Fable 5.2. Because each score combines a model with a particular agent tool, the result compares those setups as packaged, not the models alone. Google's own launch claims, and the access details, are in its announcement.

Key takeaways · AI-distilled
  • Artificial Analysis finds Neon uses about 15% more output tokens per task than Gemini 4 Carbon, so its lower cost comes entirely from 25% lower token prices ($3/$15 vs $4/$20), and its ~40% shorter task time from higher output speed (~110 vs ~57 tokens/s).
  • In Antigravity CLI, Neon (xhigh) averaged $6.86 per task on the Coding Agent Index versus $12.61 for Claude Opus 5.5 (max) and $15.47 for Claude Fable 5.2 (xhigh), while scoring 59 against their 58.
  • Neon's $3/$15 list price is 50% above Gemini 4 Carbon's discounted $2/$10, which ends November 30, and 25% below Carbon's standard $4/$20. Artificial Analysis's cost comparison uses Carbon's standard pricing.
  • On AA-Briefcase, Neon (xhigh) led Presentation Quality at 1771 Elo, 45 points ahead of Claude Sonnet 5.5 (max). On GDP.pdf, Neon at high effort scored 33.4%, above its own xhigh setting at 28.0%.
Terms in this piece · Glossary
  • token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
  • AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Read the source artificialanalysis.ai
More from Artificial Analysis articles
Recommended reads
Comments

Checking sign-in…

Loading comments…