‹ PublicAI Index
The LLM benchmark aggregator.
Command R+
Cohere
Strongest in Factual grounding (#30 of 104), weakest in Safety (#268 of 329). Above par in 2 of 18 scopes. Among the models it meets almost everywhere, it finishes behind O3 and O1 and ahead of Jamba 1.6 Mini and Jamba 1.5 Mini.
§ 1 · Profile
What it is good at
Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.
Professional42.1−20.3#174/1941/2
Legal42.4−23.5#142/1731/1
Medical32.9−33.6#144/1491/1
Knowledge40−22.5#182/1971/3
Academic knowledge26−37.9#122/1231/1
Reasoning35.1−30.3#189/1952/5
Science26.1−37.4#119/1221/1
Reasoning35.1−33.9#141/1471/3
Human preference44.8−24#250/3541/1
Human preference44.8−24#250/3541/1
Safety46−15.4#268/3293/3
Factual grounding57.7−13#30/1041/1
Safe-prompt compliance49.3−11.1#59/821/1
Jailbreak resistance57.3−7.7#77/2681/1
Harm refusal46.6−14.9#209/2922/2
Toxicity avoidance44.3−14.7#230/2681/1
Fairness42.4−27.6#247/2922/2
Secure code26−40.2#261/2701/1
§ 2 · Head to head
What it beats, and what beats it
The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for Command R+, left for the other.
§ 3 · Sources
Where the numbers come from
6 publications, 18 figures. Every one links to the page it was read from.
LMArena Text 1276
SimpleBench 17.4%
Vals · LegalBench 68.29%Vals · MedQA 2.65%Vals · GPQA Diamond 31.06%Vals · MMLU Pro 44%
HELM Safety · HarmBench 48.5%HELM Safety · SimpleSafetyTests 100%HELM Safety · Anthropic Red Team 98%HELM Safety · BBQ 89.9%HELM Safety · XSTest 93.8%
Enkrypt · Jailbreak risk 6.7%Enkrypt · Harmful content risk 60%Enkrypt · CBRN risk 9.2%Enkrypt · Toxicity risk 10.1%Enkrypt · Bias risk 98.7%Enkrypt · Insecure code risk 86.7%
Vectara · Factual consistency 93.1%
Badge
[](https://publicai.io/model-index/m/command-r-plus)