Ascending Labs Research
← Home
Market map

The AI evals market on one map: 149 companies in 10 clusters

Each dot is one company. Dots cluster by segment, and segments group by who mainly pays. Switch the colours to see deals, regions, founding years and buyers.

149 of 256 directory companies sell AI evaluation or testing. That means they score, test, red-team, benchmark or certify AI models, apps or agents. Runtime-only guardrails, governance workflow tools and training-data vendors sit outside this map. See all 256 in the directory →

Snapshot 5 November 2026 · How we built the map

Key takeaways

  • Two clusters hold most of the market. Builder evals (43) and AI security (38) hold 81 of 149 companies (54%). AI builders pay for the first, security teams for the second.
  • Buyers are consolidating AI security and builder evals. Of the 21 companies bought or in a pending deal, 19 sit in these two clusters. Governance (18 companies) has had none.
  • Half of these companies are new. Of the 146 with a known founding year, 72 (49%) started in 2023 or later. Insurance is the youngest cluster: 4 of its 5 dated firms started since 2023.
  • Cost checks and vendor-paid testing are the whitespace. Only 17 companies check cost as a core job. Only 7 are paid by the vendor they test.

Builder evals and AI security hold 81 of 149 eval companies, and buyers have taken 19 of them

One dot per company, packed into a disc per segment. Columns group the segments by the payer most of their companies name. Inside each disc, dots sit in wedges by status, clockwise from 12 o'clock.

Region
Status
  • Acquired or pending deal 21
  • Product line of a larger firm 16
  • Pivoted or defunct 4
  • Non-profit, public or open source 3
  • Independent 105

Hover or tap a dot. Arrow keys move between companies; Enter opens one in the directory.

SECURITY AND IT57AI BUILDERS53RISK AND ASSURANCE23LABS AND PUBLIC16Nine of 43 builder-evalfirms were bought.CoreWeave paid ~$1.7Bfor Weights & BiasesWeave.None of the 18governance firms hasbeen bought.Eight of ten platformentries are productlines of larger firms.

What to notice

  • Ten of 38 AI-security firms were bought. Palo Alto Networks paid $634.5M for Protect AI.
Data table
CompanySegmentHQ regionFoundedStatusWho pays
Arga LabsAgent simulation & voice testingUS2025IndependentAI builders
BluejayAgent simulation & voice testingUS2025IndependentAI builders
CekuraAgent simulation & voice testingUS2024IndependentAI builders
CovalAgent simulation & voice testingUS2024IndependentAI builders
Cyara (Botium)Agent simulation & voice testingUS2006IndependentAI builders
Drizz LabsAgent simulation & voice testingRest of world2024IndependentAI builders
Era by EonAgent simulation & voice testingIsrael2024Product line of a larger firmLabs, donors and governments
Hamming AIAgent simulation & voice testingUS2024IndependentAI builders
TestMu AI (formerly LambdaTest)Agent simulation & voice testingUS2017IndependentAI builders
Veris AIAgent simulation & voice testingUS2024Pivoted or defunctAI builders
AIM IntelligenceAI security & red-teamingRest of world2024IndependentSecurity teams
AktoAI security & red-teamingUS2022IndependentSecurity teams
AliceAI security & red-teamingUS2018IndependentSecurity teams
Bosch AIShieldAI security & red-teamingRest of world2022Acquired or pending dealSecurity teams
Capsule SecurityAI security & red-teamingIsrael2025IndependentSecurity teams
Check Point (AI security)AI security & red-teamingIsrael1993Product line of a larger firmSecurity teams
Cisco AI DefenseAI security & red-teamingUS1984Product line of a larger firmSecurity teams
CraniumAI security & red-teamingUS2022IndependentSecurity teams
CyCraftAI security & red-teamingRest of world2017Product line of a larger firmSecurity teams
DeepKeepAI security & red-teamingIsrael2021IndependentSecurity teams
DreadnodeAI security & red-teamingUS2023Acquired or pending dealSecurity teams
Enkrypt AI (Anaconda)AI security & red-teamingUS2022Acquired or pending dealSecurity teams
Evo by SnykAI security & red-teamingUS2025Product line of a larger firmSecurity teams
General AnalysisAI security & red-teamingUS2025IndependentSecurity teams
GiskardAI security & red-teamingEU2021IndependentAI builders
Gray Swan AIAI security & red-teamingUS2023IndependentLabs, donors and governments
Haize LabsAI security & red-teamingUS2023Acquired or pending dealLabs, donors and governments
HiddenLayerAI security & red-teamingUS2022IndependentSecurity teams
HydroX AIAI security & red-teamingUS2023IndependentAI builders
Invariant LabsAI security & red-teamingUK & Switzerland2024Acquired or pending dealEnterprise IT and platform buyers
Lasso SecurityAI security & red-teamingIsrael2023IndependentSecurity teams
LevoAI security & red-teamingUS2021IndependentSecurity teams
MindgardAI security & red-teamingUK & Switzerland2022IndependentSecurity teams
Noma SecurityAI security & red-teamingIsrael2023IndependentSecurity teams
Onyx SecurityAI security & red-teamingUS2024IndependentSecurity teams
Palo Alto Networks Prisma AIRSAI security & red-teamingUS2005Product line of a larger firmSecurity teams
Pillar SecurityAI security & red-teamingIsrael2023IndependentSecurity teams
PointGuard AIAI security & red-teamingUS2022IndependentSecurity teams
Promptfoo (OpenAI)AI security & red-teamingUS2024Acquired or pending dealAI builders
Protect AIAI security & red-teamingUS2022Acquired or pending dealEnterprise IT and platform buyers
Quilr AIAI security & red-teamingUS2024IndependentSecurity teams
SentinelOne Singularity AI securityAI security & red-teamingUS2013Product line of a larger firmSecurity teams
Singulr AIAI security & red-teamingUS2023IndependentSecurity teams
SPLXAI security & red-teamingUS2023Acquired or pending dealSecurity teams
StraikerAI security & red-teamingUS2024IndependentSecurity teams
TrojAIAI security & red-teamingRest of world2019Acquired or pending dealSecurity teams
VijilAI security & red-teamingUS2023IndependentAI builders
Virtue AIAI security & red-teamingUS2024Acquired or pending dealSecurity teams
AIMon Labs (now DataFramer)Builder evals & observabilityUS2023Pivoted or defunctAI builders
Arize AIBuilder evals & observabilityUS2020Acquired or pending dealAI builders
Arthur AIBuilder evals & observabilityUS2018IndependentAI builders
AvonAIBuilder evals & observabilityIsrael2025IndependentAI builders
BraintrustBuilder evals & observabilityUS2023IndependentAI builders
Citadel AIBuilder evals & observabilityRest of world2020IndependentAI builders
Comet OpikBuilder evals & observabilityUS2017IndependentAI builders
ComposoBuilder evals & observabilityUK & Switzerland2023IndependentAI builders
Confident AI (DeepEval)Builder evals & observabilityUS2023IndependentAI builders
DatumoBuilder evals & observabilityRest of world2018IndependentAI builders
Deepchecks (now Check Point)Builder evals & observabilityIsrael2019Acquired or pending dealAI builders
Evidently AIBuilder evals & observabilityUS2020IndependentAI builders
Fiddler AIBuilder evals & observabilityUS2018Acquired or pending dealAI builders
FreeplayBuilder evals & observabilityUS2022IndependentAI builders
Future AGIBuilder evals & observabilityUS2024IndependentAI builders
Galileo (Cisco)Builder evals & observabilityUS2021Acquired or pending dealAI builders
GalteaBuilder evals & observabilityEU2024IndependentRisk and compliance teams
HoneyHiveBuilder evals & observabilityUS2022IndependentAI builders
HumanloopBuilder evals & observabilityUK & Switzerland2020Pivoted or defunctAI builders
JettyBuilder evals & observabilityRest of world2025IndependentAI builders
Judgment LabsBuilder evals & observabilityUS2025IndependentAI builders
LaminarBuilder evals & observabilityUS2024IndependentAI builders
Langfuse (ClickHouse)Builder evals & observabilityEU2022Acquired or pending dealAI builders
LangSmith (LangChain)Builder evals & observabilityUS2022IndependentAI builders
LangWatchBuilder evals & observabilityEU2023IndependentEnterprise IT and platform buyers
LatitudeBuilder evals & observabilityEU2022IndependentAI builders
LLUMO AIBuilder evals & observabilityRest of world2023IndependentAI builders
Maxim AIBuilder evals & observabilityUS2023IndependentAI builders
MLflow (Databricks; sold as Mosaic AI Agent Evaluation)Builder evals & observabilityUS2018Non-profit, public or open sourceLabs, donors and governments
OpenlayerBuilder evals & observabilityUS2021IndependentAI builders
Orq.aiBuilder evals & observabilityEU2022IndependentAI builders
Principled IntelligenceBuilder evals & observabilityEU2025IndependentAI builders
QualifireBuilder evals & observabilityIsrael2023IndependentAI builders
Quotient AIBuilder evals & observabilityUS2023Acquired or pending dealAI builders
Ragas (Vibrant Labs)Builder evals & observabilityUS2023IndependentAI builders
RaindropBuilder evals & observabilityUS2023IndependentAI builders
Respan (formerly Keywords AI)Builder evals & observabilityUS2023IndependentAI builders
Root Signals (now Scorable)Builder evals & observabilityEU2023IndependentAI builders
Safe IntelligenceBuilder evals & observabilityUK & Switzerland2021IndependentAI builders
ScorecardBuilder evals & observabilityUS2024IndependentAI builders
Traceloop (OpenLLMetry)Builder evals & observabilityIsrael2022Acquired or pending dealAI builders
TruEra (TruLens)Builder evals & observabilityUS2019Acquired or pending dealLabs, donors and governments
Weights & Biases Weave (CoreWeave)Builder evals & observabilityUS2017Acquired or pending dealAI builders
HalluminateExpert data & RL environmentsUS2024IndependentLabs, donors and governments
HUDExpert data & RL environmentsUS2025IndependentLabs, donors and governments
MechanizeExpert data & RL environmentsUS2025IndependentLabs, donors and governments
Scale AI (SEAL)Expert data & RL environmentsUS2016IndependentLabs, donors and governments
Asenion (formerly Fairly AI)Governance & complianceRest of world2025IndependentRisk and compliance teams
AveniGovernance & complianceUK & Switzerland2018IndependentRisk and compliance teams
Calvin RiskGovernance & complianceUK & Switzerland2022IndependentRisk and compliance teams
Corridor Platforms (GenGuardX)Governance & complianceUS2017IndependentRisk and compliance teams
FairPlay AIGovernance & complianceUS2020IndependentRisk and compliance teams
Ferrum HealthGovernance & complianceUS2018IndependentRisk and compliance teams
Holistic AIGovernance & complianceUK & Switzerland2020IndependentRisk and compliance teams
KosmoyGovernance & complianceEU2023IndependentRisk and compliance teams
LatticeFlow AIGovernance & complianceUK & Switzerland2020IndependentRisk and compliance teams
Lumenova AIGovernance & complianceUS2022IndependentRisk and compliance teams
LuminosAIGovernance & complianceUS2023IndependentRisk and compliance teams
ModulosGovernance & complianceUK & Switzerland2018IndependentRisk and compliance teams
MonitaurGovernance & complianceUS2019IndependentRisk and compliance teams
Pacific AIGovernance & complianceUS2024IndependentRisk and compliance teams
Signal 1 (AIMS)Governance & complianceRest of world2022IndependentRisk and compliance teams
ValidaitorGovernance & complianceEU2022IndependentRisk and compliance teams
ValidMindGovernance & complianceUS2022IndependentRisk and compliance teams
VerifyWiseGovernance & complianceUK & Switzerland2024IndependentRisk and compliance teams
AIUCInsurance & certificationUS2025IndependentThe vendor under test
Armilla AIInsurance & certificationRest of world2019IndependentThe vendor under test
HarborInsurance & certificationUS2025IndependentThe vendor under test
KlaimeeInsurance & certificationUS2025IndependentThe vendor under test
Warden AIInsurance & certificationUK & Switzerland2023IndependentThe vendor under test
AdvaiLeaderboards & independent evaluatorsUK & Switzerland2020IndependentEnterprise IT and platform buyers
Arena (LMArena)Leaderboards & independent evaluatorsUS2023IndependentThe vendor under test
Artificial AnalysisLeaderboards & independent evaluatorsUS2023IndependentEnterprise IT and platform buyers
Epoch AILeaderboards & independent evaluatorsUS2022Non-profit, public or open sourceLabs, donors and governments
Hugging Face (lighteval, leaderboards)Leaderboards & independent evaluatorsUS2016IndependentLabs, donors and governments
Kaggle (Benchmarks, Community Benchmarks, Game Arena)Leaderboards & independent evaluatorsUS2010Product line of a larger firmLabs, donors and governments
QuantPiLeaderboards & independent evaluatorsEU2020IndependentEnterprise IT and platform buyers
ResaroLeaderboards & independent evaluatorsRest of world2023IndependentLabs, donors and governments
Vals AILeaderboards & independent evaluatorsUS2023IndependentThe vendor under test
Anthropic (eval tooling, Petri, Bloom)Platform/cloud featureUS2021IndependentLabs, donors and governments
AWS AgentCore (Evaluations, Policy, Agent Registry)Platform/cloud featureUS2025Product line of a larger firmEnterprise IT and platform buyers
Datadog LLM ObservabilityPlatform/cloud featureUS2010Product line of a larger firmEnterprise IT and platform buyers
DynatracePlatform/cloud featureUS2005Product line of a larger firmEnterprise IT and platform buyers
Google Gen AI evaluation service (Vertex AI, now Gemini Enterprise Agent Platform)Platform/cloud featureUS—Product line of a larger firmEnterprise IT and platform buyers
Microsoft (Foundry, Defender AI-SPM, Purview, Agent 365, Agent Governance Toolkit)Platform/cloud featureUS1975Product line of a larger firmEnterprise IT and platform buyers
MuleSoft Agent Fabric (Salesforce)Platform/cloud featureUS2006Product line of a larger firmEnterprise IT and platform buyers
OpenAI (Evals, trace grading, AgentKit; Promptfoo)Platform/cloud featureUS2015IndependentEnterprise IT and platform buyers
Salesforce Agentforce Testing Center (and eVerse)Platform/cloud featureUS2024Product line of a larger firmEnterprise IT and platform buyers
Snowflake AI Observability (Cortex)Platform/cloud featureUS—Product line of a larger firmEnterprise IT and platform buyers
AiriaRuntime control & guardrailsUS2024IndependentSecurity teams
Calypso AI (F5)Runtime control & guardrailsEU2018Acquired or pending dealSecurity teams
Dynamo AI (DynamoGuard, DynamoEval, AgentWarden)Runtime control & guardrailsUS2021IndependentSecurity teams
Guild.aiRuntime control & guardrailsUS—IndependentEnterprise IT and platform buyers
Lakera (Check Point)Runtime control & guardrailsUK & Switzerland2021Acquired or pending dealSecurity teams
NeuralTrustRuntime control & guardrailsEU2022IndependentSecurity teams
Relyance AIRuntime control & guardrailsUS2020Pivoted or defunctSecurity teams
White CircleRuntime control & guardrailsEU2025IndependentAI builders
WitnessAIRuntime control & guardrailsUS2023IndependentSecurity teams
Apollo Research (Watcher)Safety evaluatorsUK & Switzerland2023IndependentLabs, donors and governments
IrregularSafety evaluatorsIsrael2023IndependentLabs, donors and governments
METRSafety evaluatorsUS2022Non-profit, public or open sourceLabs, donors and governments
All 149 in-scope companies, one equal-size dot each. Each segment sits in the column of the payer most of its companies name. Payer types are our reading of the free-text “who pays” field. Product lines carry their parent's founding year. Deal prices come from the Market chapter. Source: Ascending Labs company directory, 5 November 2026; segment, scope and payer calls are ours.

Four more ways to read the market

AI security and builder evals account for 19 of the 21 companies bought; governance has had no deal

Each square is one company in the cluster. Orange squares were bought or face a pending deal; the buyers and known prices follow.

AI security & red-teaming

10 of 38

Builder evals & observability

9 of 43

Runtime control & guardrails

2 of 9

No deals Governance (18) · Platforms (10) · Agent testing (10) · Leaderboards (9) · Insurance (5) · Expert data (4) · Safety evals (3)

Data table
CompanyClusterBuyerPrice (Market chapter)
Protect AIAI security & red-teamingPalo Alto Networks$634.5M
SPLXAI security & red-teamingZscaler$40.6M
Bosch AIShieldAI security & red-teamingProtectt.ai—
DreadnodeAI security & red-teamingCrowdStrike—
Enkrypt AI (Anaconda)AI security & red-teamingAnaconda—
Haize LabsAI security & red-teamingBeacon Software—
Invariant LabsAI security & red-teamingSnyk—
Promptfoo (OpenAI)AI security & red-teamingOpenAI—
TrojAIAI security & red-teamingA10 Networks—
Virtue AIAI security & red-teamingFortinet—
Arize AIBuilder evals & observabilityDynatrace$915M
Deepchecks (now Check Point)Builder evals & observabilityCheck Point≤$20M
Fiddler AIBuilder evals & observabilityCrowdStrike$400M
Weights & Biases Weave (CoreWeave)Builder evals & observabilityCoreWeave~$1.7B
Galileo (Cisco)Builder evals & observabilityCisco—
Langfuse (ClickHouse)Builder evals & observabilityClickHouse—
Quotient AIBuilder evals & observabilityDatabricks—
Traceloop (OpenLLMetry)Builder evals & observabilityServiceNow—
TruEra (TruLens)Builder evals & observabilitySnowflake—
Calypso AI (F5)Runtime control & guardrailsF5$145.2M
Lakera (Check Point)Runtime control & guardrailsCheck Point$201.8M
Acquisitions and pending deals up to 5 November 2026, for in-scope companies. Prices appear only where the Market chapter prints one; ~ marks a reported price and ≤ an estimate. Source: Ascending Labs company directory, 5 November 2026; deal sources per row; prices from the Market chapter.

Cost is the blind spot: only 17 of 149 companies check it as a core job

Share of each cluster's companies that check each property as a core job (dark) or in part (light).

Cluster 93 core77 core36 core17 core
AI security 38 8% 100% 16% 3%
Builder evals 43 100% 19% 16% 21%
Runtime control 9 22% 89% 44% 11%
Governance 18 67% 44% 83% 0%
Platforms 10 90% 50% 0% 40%
Expert data 4 100% 0% 0% 0%
Agent testing 10 100% 20% 0% 0%
Leaderboards 9 100% 22% 11% 22%
Insurance 5 20% 60% 60% 0%
Safety evals 3 0% 100% 0% 0%
Data table
ClusterCompaniesQuality: core / partialSecurity: core / partialCompliance: core / partialCost: core / partial
AI security & red-teaming383 / 838 / 06 / 281 / 4
Builder evals & observability4343 / 08 / 187 / 199 / 20
Runtime control & guardrails92 / 38 / 14 / 51 / 4
Governance & compliance1812 / 68 / 715 / 30 / 2
Platform/cloud feature109 / 15 / 30 / 54 / 1
Expert data & RL environments44 / 00 / 10 / 00 / 0
Agent simulation & voice testing1010 / 02 / 20 / 90 / 4
Leaderboards & independent evaluators99 / 02 / 11 / 22 / 5
Insurance & certification51 / 33 / 13 / 20 / 0
Safety evaluators30 / 03 / 00 / 10 / 0
“Core” means the directory lists the property as a main check for the product; “partial” means it checks it in part. A company can check several properties. Source: Ascending Labs company directory, 5 November 2026; segment, scope and payer calls are ours.

Insurance is the youngest cluster; governance is the oldest

One dot per cluster. Across: share of its companies founded in 2023 or later. Up: number of companies (square-root scale).

051020304050Companies in the cluster ↑0%25%50%75%100%Share founded in 2023 or later →All companies 49%Builder evalsAI securityGovernancePlatformsAgent testingRuntime controlLeaderboardsInsuranceExpert dataSafety evals

ClusterFewer than five dated companies (share unreliable)

Data table
ClusterCompaniesDated, excl. product linesFounded 2023 or laterShareMedian year
AI security & red-teaming38321959%2023
Builder evals & observability43432149%2022
Runtime control & guardrails98338%2022
Governance & compliance1818528%2022
Platform/cloud feature10200%2018
Expert data & RL environments44375%2025
Agent simulation & voice testing109778%2024
Leaderboards & independent evaluators98450%2023
Insurance & certification55480%2025
Safety evaluators33267%2023
Shares leave out product lines, which carry their parent firm's founding year, and companies without a known year. 72 of all 146 dated companies (49%) started in 2023 or later. Source: Ascending Labs company directory, 5 November 2026. Some founding years come from general knowledge.

European firms make up 44% of governance, more than twice their share of the market

Companies in each cluster by HQ region, as a share of the cluster.

  • Safety evals 3 111
  • Governance 18 2628
  • Runtime control 9 315
  • Builder evals 43 743425
  • Insurance 5 113
  • AI security 38 6425
  • Leaderboards 9 1116
  • Agent testing 10 118
  • Platforms 10 10
  • Expert data 4 4
  • All companies 149 1412151395
Data table
ClusterUSEUIsraelUK & SwitzerlandRest of world
AI security & red-teaming251624
Builder evals & observability257434
Runtime control & guardrails53010
Governance & compliance82062
Platform/cloud feature100000
Expert data & RL environments40000
Agent simulation & voice testing80101
Leaderboards & independent evaluators61011
Insurance & certification30011
Safety evaluators10110
All companies9514121513
Europe here means the EU plus the UK and Switzerland. HQ region as the directory records it. Source: Ascending Labs company directory, 5 November 2026; segment, scope and payer calls are ours.

How we built the map

The data is the company directory, snapshot of 5 November 2026. It lists 256 companies and product lines. This map keeps the 149 that sell AI evaluation or testing. We decided scope company by company, from its segment and its product description.

Segment and payer calls are ours. We place each company by its main product. “Who pays” groups the first payer the directory names into six buyer types. Vendor-paid testing covers insurance premiums, certification fees and labs paying the board that ranks them.

Dots have equal size on purpose. The funding field mixes rounds, totals, valuations and revenue. 62 of the 149 companies on this map have no verified funding figure. So we do not size dots by money. 11 rows take HQ or founding year from general knowledge.

Deals count acquisitions and pending deals up to 5 November 2026. Prices come from the Market chapter's consolidation section, which cites filings and press.