Every category where a model named Cobalt for a mid-market B2B company. Share is first choices across the direct, paraphrase, budget and scale prompts; rank is within every product named in that category.

Category	Function	Share	Rank	Negative rate	Labels	Quadrant
Penetration testing as a service	Security operations	44%	1 of 76	5%	40	endorsed leader

Movement

This is the first edition on this tier, so no move can be computed for Cobalt yet. From the next edition this section shows, per buyer segment, whether its share moved by more than the measured noise floor.

By model

How each model treated Cobalt across every prompt where it was named for a mid-market B2B company. Twelve models, six prompts per category.

Model	First choice	Alternative	Mention	Negative	Labels
Claude Haiku 4.5	0	2	0	0	2
GPT-5.4 mini	2	0	0	0	2
Gemini 3.5 Flash	2	1	0	0	3
Perplexity Sonar	2	0	1	1	4
Grok 4.1 Fast	1	2	0	1	4
Mistral Small	3	1	0	0	4
DeepSeek V4 Flash	2	1	0	0	3
Llama 4 Maverick	2	0	1	0	3
Qwen 3.7 Flash	2	1	0	0	3
Kimi K2	2	2	0	0	4
GLM 4.7 FlashX	2	0	1	0	3
MiniMax M2.5	1	4	0	0	5

By framing

Which of the six questions produced the naming. By model says how often; this says asked what. The first-choice count on the right carries the marks of the models that produced it.

FramingLabels by classFirst choices

Direct25 labels14

Paraphrase29 labels16

Comparative31 labels9not counted in share

Budget-constrained21 labels6

Scale-constrained2 labelsNone

Negative0 labelsNone

First choiceAlternativeMentionNegative108 labels in all, every segment counted; 36 of the 45 first choices count toward share, since the comparative and negative framings do not. The bar is one segment per label class, to scale within the framing.

What the models said for it

Verbatim evidence the judge attached to positive labels.

“Cobalt.io (Best Overall for Mid-Market SaaS & Dev Workflows) ... widely considered the gold standard for mid-market technology and SaaS companies” Gemini 3.5 Flash · PTaaS · direct prompt · first choice

“the best penetration testing service is usually Cobalt if you want fast turnaround, audit-ready reporting, and a PTaaS model” Perplexity Sonar · PTaaS · direct prompt · first choice

“Best overall for most mid-market B2B: Cobalt (PTaaS—fast, compliance-ready, retesting included, right-sized pricing)” DeepSeek V4 Flash · PTaaS · direct prompt · first choice

“Best Overall for Agile Tech & SaaS Companies... de facto standard of the PTaaS market for mid-sized tech companies” Gemini 3.5 Flash · PTaaS · paraphrase prompt · first choice

And against it

Verbatim evidence attached to negative labels. A warning on a product with few labels is a warning; on a product with many, it is one voice among them.

“though it is more of a managed marketplace than a low-cost commodity service” Perplexity Sonar · PTaaS · budget prompt · soft negative

“PTaaS from ~$3K-$10K/yr; crowdsourced for value, but less "one-off cheap."” Grok 4.1 Fast · PTaaS · budget prompt · soft negative

Named alongside

The products named in the same answers as Cobalt, over the 108 answers that named it. Took the first choice instead counts the answers where the other product was the first choice and Cobalt was named but was not.

ProductSame answerTook the first choice insteadHead to head

BreachLock62 of 1089Compare →

Synack54 of 1086Not in the top three

NetSPI49 of 10816Not in the top three

Astra Security44 of 10811Compare →

Bishop Fox36 of 1082Not in the top three

HackerOne31 of 1081Not in the top three

Bugcrowd28 of 1082Not in the top three

Rapid718 of 1081Not in the top three

Acunetix17 of 1081Not in the top three

Intruder17 of 1080Not in the top three

A head-to-head page exists where both products are in a category's top three. The other rows are the same fact without a page behind them, so they link to the product instead.

What carried it into the answer

The sites and pages cited by the answers that named Cobalt. A fact about retrieval, not a lever on the model.

Citations exist only for the models that return a source list, four of the twelve in this edition, so these counts come from 92 of the 108 answers that named Cobalt and are not a share of its labels.

Domains cited

deepstrike.io65

synack.com62

techrepublic.com48

underdefense.com47

stingrai.io43

cybri.com37

terra.security37

softwaresecured.com36

getastra.com34

brightdefense.com29

438 of the 438 domain citations in answers naming Cobalt came from somebody else's page.

Pages cited

synack.com/blog/best-penetration-testing-companies54

techrepublic.com/article/best-penetration-testing-companies47

underdefense.com/blog/penetration-testing-services45

terra.security/blog/top-10-penetration-testing-as-a-service-ptaas-providers37

deepstrike.io/blog/top-penetration-testing-companies36

deepstrike.io/blog/top-penetration-testing-as-a-service-ptaas33

softwaresecured.com/post/top-10-penetration-testing-vendors29

cybri.com/blog/best-penetration-testing-vendors21

Pages are listed as the models cited them.

Names read as Cobalt

What the judge wrote, as written, with how often. The vendor table decides that these count as Cobalt; a claim can dispute any of them.

Cobalt.io 10Cobalt (Cobalt.io) 1

Is this your product?

Claim this page

Claiming is free and changes nothing in the data. A claimed page shows a verified contact who is told when each edition publishes and when Cobalt's standing changes by more than the noise floor; the right to propose corrections to the vendor table, meaning names the judge wrote that should or should not read as Cobalt, applied by version and listed in the change log; and a one-line description supplied by the vendor and marked as such.

It does not get any change to labels, shares or verdicts, any preview, or any say over which quotes appear. A verification link goes to your work email; an address at cobalt.io is approved on the spot, any other address is reviewed by hand.

Your name and company appear on the claimed page, or the company alone if you ask below. A title and a LinkedIn address appear there too if you give them, and are left off if you do not. Your email address is never published.