Google's Gemini 4 Carbon edges past Claude Opus 5.5 on AI benchmarks
Access is starting slowly with cyber defenders, and the hallucination gains are company claims that the benchmarks show are still far from consistent reliability.
Google has released Gemini 4 Carbon, which Artificial Analysis ranks first on its Intelligence Index V5 with a score of 59, one point ahead of Claude Opus 5.5. It also leads the AA-Omniscience test, which measures knowledge recall and hallucination, with a score of 54, though even that top result is far from consistent reliability. Its cost per task of $7.76 is slightly below Gemini 3.8 Flash and several dollars under Claude Opus 5.5, yet it remains pricier than several rivals. Access is being rolled out slowly, starting with cyber defenders through the Fairwind program, while Google says it is working with the U.S. government on pre-release testing. Google also claims Carbon reduces unsupported claims at twice the rate of its predecessor, Gemini 4 Argon, and cuts model overreach by 26%.
Why it matters
The lead on the Intelligence Index is a single point, and the top score on the hallucination-focused test is still only 54, so the model is better at recall and self-restraint by Google's measures but not dependable. For users, the practical shift is that a rival now leads on the closest measure of accuracy while remaining costlier than several competitors, and a slow, staged rollout means most people will not be able to test those claims themselves for some time.
Matters or hype?
Does this story matter, or is it hype? Vote before you see what everyone else said.