[whatmodel.sh](../../README.md) / Gemini 4 Carbon # Gemini 4 Carbon Google · released 2026-10 · closed weights **score 83.4** ±1.3 · #4 of 732 · from 3 benchmark results across 1 source Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only. ## What it's good at ● marks Gemini 4 Carbon among every model on that task or test, worst to best. 
task            vs the others        rank  counts  tests
agent tasks     ▁▁▃▆█▄▅▅▄▄▂●   #2 of 313*     48%      1
coding          ▁▁▁▃▅▇█▇▅▄▃●   #3 of 469*     33%      1
chat & writing  ▁▁▁▃▆▄▆█▆▇▇●   #1 of 402*     19%      1
agent tasks · #2 of 313 · 48% of its score Beats [GPT-6 Astra](gpt-6-astra.md), [GPT-6.1 Terra](gpt-6-1-terra.md) here, though they rank higher overall. 
test                                    result  vs the others in its run  difficulty      counts  why
Arena Agent (October 2026 aggrega…  0.1604 IPS  ▁▁▁▃▃▂█▁▂▂●▁ beats 58/59  avg -0.379 IPS     48%  sharp
coding · #3 of 469 · 33% of its score Beats [GPT-6.1 Terra](gpt-6-1-terra.md) here, though they rank higher overall. 
test                       result  vs the others in its run    difficulty  counts  why
Arena WebDev (Code Arena)  1768.2  ▁▁▃▂▅█▃▆▅▅▁● beats 126/129  avg 981        33%  sharp
chat & writing · #1 of 402 · 19% of its score Beats [GPT-6 Astra](gpt-6-astra.md), [Claude Opus 5.5](claude-opus-5-5.md) here, though they rank higher overall. 
test            result  vs the others in its run    difficulty  counts  why
LMArena (text)  1527.5  ▁▁▁▂▃▃▃▆▆▇█● beats 385/385  avg 1260       19%  blunt
other tests (2) 
test                            result  vs the others in its run  counts  why
SEC-Bench Pro (pass@1) (maker)   88.2%  ▄·▄·▄·█▄···● beats 6/6         —  not counted yet: no other model in the …
ExploitGym (pass@1) (maker)      54.8%  ▄·▄··█·▄▄··● beats 6/6         —  not counted yet: no other model in the …
## How the score was worked out Each result implies a score on its own, after adjusting for how hard that test is and who ran it. The model's score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its *counts* column above; here it is added up by who ran the tests. [Method](../method.md) 
source                         results  share of the score
LMArena leaderboard dataset          3   99% ▇▇▇▇▇▇▇▇▇▇▇▇
prior: pull toward the middle             1% ▇
weighted average                        = 83.4
Licences: LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/))
Not counted (2) 
source                  test                    score  why
Google DeepMind Gemin…  SEC-Bench Pro (pass@1)  88.2%  too few models have it yet
Google DeepMind Gemin…  ExploitGym (pass@1)     54.8%  too few models have it yet
## How to run it No verified API or local download yet. [interactive version](https://whatmodel.sh/model/gemini-4-carbon) · [JSON](https://whatmodel.sh/api?format=json&all=1)