Gemini 4 Carbon is now available

287 points · 159 comments on HN · read original →

Points and comments are a snapshot, not live.

Google releases Gemini 4 Carbon, its strongest frontier model yet.

Gemini 4 Carbon is the strongest model in Google's Gemini 4 lineup, focused on cybersecurity, multi-step coding, and complex scientific reasoning. It ranks #1 on CursorBench 4.0, Perplexity's WANDR, and Zapier's AutomationBench at more than 2x lower cost per task than the next best model. Carbon offers a new /swarm feature for multi-agent coordination and matches Argon's pricing ($2 per million input, $10 per million output). It is available via AI Studio, Gemini for business, and to AI Ultra subscribers in Antigravity, the Gemini app, and AI Mode in Search. The model features enhanced safety guardrails, blocking 99% of attack trajectories in red-teaming tests.

What commenters are saying

Users are frustrated with Google's rollout and tier structure. Many on the $20/mo Pro plan report lacking access to Carbon and even Argon, while others access them with different subscriptions. Several complain that Pro limits are so restrictive they can only complete a few tasks per week. A few commenters note that the $100/mo Ultra plan provides usable quotas. The thread also highlights confusion over multiple different 'Pro' tiers (pixel promos, Jio, education) with inconsistent model access, and general sentiment that Google's product offering is confusing compared to OpenAI or Anthropic.