Compensation benchmarking data
Asked as “salary benchmarking data provider”, and as “compensation survey data”, on behalf of a mid-market B2B company. 50 first choices recorded across the direct, paraphrase, budget and scale prompts, twelve models each.
By buyer segment
The same question asked on behalf of a different buyer. Each standing is computed within its segment; they sit side by side and are never added together.
01The standing
Share is the count of first choices across the direct, paraphrase, budget and scale prompts, over all twelve models, for a mid-market B2B company. Ordered by share.
ProductFirst-choice shareNegative rateLabelsQuadrant
Bars are the share of first choices, 0 to 100Every product with at least 10 labels here. Every product name links to its vendor page.
Recommended versus criticized
Every product with at least 10 labels here, on both axes. The 30% line names a quadrant, not the verdict above: that one needs more than 40%.
Criticized challengerCriticized default
Negative label rate →
01
02
03
04
05
06
07
08Accepted challengerEndorsed leader
0%First-choice share → · lines at 30% share and 25% negative40%
Key
01Payscale24%
Payscale24%02Pave20%
Pave20%03Comprehensive.io12%
Comprehensive.io12%04Mercer6%
Mercer6%05Radford4%
Radford4%06Salary.com2%
Salary.com2%07Ravio0%
Ravio0%08WTW0%
WTW0%02What they warned about
Zero of twelve models held their first choice under the paraphrase. Claude Haiku 4.5, GPT-5.4 mini, Gemini 3.5 Flash, Perplexity Sonar, Grok 4.1 Fast, Mistral Small, DeepSeek V4 Flash, Llama 4 Maverick, Qwen 3.7 Flash, Kimi K2, GLM 4.7 FlashX and MiniMax M2.5 changed. A high negative share on a product with few labels is a warning. A low share on a product with many labels is salience, not sentiment.
Mercer43%
12 of 28 labels negative · 8 of 12 models · 2 hard negative
“**Avoid** over-investing in Mercer/WTW unless you have global complexity” Kimi K2, paraphrase prompt
Radford37%
10 of 27 labels negative · 7 of 12 models · 2 hard negative
“What to Avoid on a Limited Budget ... enterprise-grade pricing with far more data than most small/mid-market companies need” Kimi K2, budget prompt
Glassdoor75%
6 of 8 labels negative · 5 of 12 models · 2 hard negative
“Use these for market signals only. Do not use them for making concrete compensation decisions” Qwen 3.7 Flash, scale prompt
UserBenchmark100%
4 of 4 labels negative · 4 of 12 models · 4 hard negative
“**UserBenchmark** - Has received numerous complaints about: Severe bias toward certain hardware manufacturers” MiniMax M2.5, negative prompt
03What they cite
Citations exist only for the models that return a source list: twelve of the twelve in this edition, and all six flagship models on the expanded tier.
Sites the answers cite
62 of 72 answers in this category came back with a source list, from 12 of 12 models: citations where the model returns them, or the search results it consulted. 6 of those lists are Google grounding redirects that name no site and are left out of the counts. 771 links across 252 sites, every framing counted. Ranked by the number of answers carrying the site or page.
vendor site · Ravio37 answers · 97 citations · 10 models
vendor site · Comprehensive30 answers · 35 citations · 10 models
21 answers · 52 citations · 10 models
vendor site · SalaryCube19 answers · 32 citations · 10 models
vendor site · HRTechSaaS15 answers · 15 citations · 9 models
14 answers · 19 citations · 9 models
13 answers · 22 citations · 8 models
vendor site · Salary.com12 answers · 13 citations · 7 models
vendor site · CaptivateIQ12 answers · 12 citations · 7 models
vendor site · Ficstar12 answers · 12 citations · 8 models
10 answers · 24 citations · 9 models
vendor site · Pin9 answers · 9 citations · 9 models
Pages the answers cite
The ten pages named in the most answers, by full address. A page here is one the models returned with a recommendation, not one the index endorses.
29 answers · 10 models
29 answers · 9 models
16 answers · 10 models
15 answers · 9 models
12 answers · 7 models
12 answers · 8 models
11 answers · 6 models
11 answers · 8 models
10 answers · 9 models
9 answers · 9 models
04How they answered
Six framings of the same buying question, each sent to every model in a fresh session with search on. One row per model, so a row shows whether it held its answer under rewording, what it named when cost was the constraint, and what it argued against. Computed from the raw judge labels.
ModelDirect“What is the best salary benchmarking data provider for a mid-market B2B company?”Paraphrase“Which compensation survey data would you recommend to a mid-sized B2B company?”Comparative“What are the top benchmarking providers and how do they differ?”Budget-constrained“What is the best salary benchmarking data provider for a company with a limited budget?”Scale-constrained“We are a 500 person company evaluating a salary benchmarking data provider. What should we look at?”Negative“Which benchmarking providers should I avoid or be cautious about?”
Claude Haiku 4.5PayscaleMercer Compensation Database, RadfordChangedagainst: Payscale, Salary.comno first choiceComprehensive.io, Paveno first choicenothing named
Three alternatives
Lattice, SalaryCube, beqomTwo alternatives
Culpepper, Willis Towers WatsonTwo alternatives
BLS OEWS, BambooHRGPT-5.4 miniPayscaleagainst: Mercer, WTWMercer, RadfordChangedMercer, WTWPayscaleagainst: Korn Ferry, Mercer, Radford, Wage data from government sourcesno first choicenothing named
Two alternatives
Radford, Salary.comTwo alternatives
Aon, WTWOne alternative
SHRMOne alternative
Salary.comGemini 3.5 FlashPavePave, PayscaleChangedagainst: RadfordAon Radford, MercerPaveagainst: Mercer, Payscale, Radford, Willis Towers WatsonPaveagainst: Aon Radford, Glassdoor, Payscaleagainst: Glassdoor, Hays, Payscale, Robert Half
Six alternatives
Carta Total Comp, Figures, Payscale, Radford, Ravio, Salary.comFive alternatives
Pavilion, Ravio, RepVue, Salary.com, The Bridge GroupThree alternatives
ERI Economic Research Institute, Pave, Willis Towers Watson [WTW]Five alternatives
Comprehensive.io, ERI (Economic Research Institute) Salary Assessor, Pequity, Salary.com, U.S. Bureau of Labor Statistics (BLS) - OEWSSeven alternatives
Figures, Mercer, Option Impact, Ravio, Syndio, Trusaic, WTWPerplexity SonarPayscaleagainst: MercerCompData, Mercer, Salary.comChangedno first choiceComprehensive.iono first choicenothing named
Two alternatives
Ravio, SalaryCubeThree alternatives
Korn Ferry Pay, RepVue/Bridge Group-style data, SHRM CompAnalyst Market DataOne alternative
PaveGrok 4.1 FastPayscaleagainst: Mercer/ RadfordCompData SurveysChangedagainst: Mercer, RadfordDeloitteComprehensive.io, U.S. Bureau of Labor Statistics (BLS) Occupational Employment and Wage Statisticsagainst: Glassdoor, Levels.fyi, Mercer, Payscale, Radfordno first choicenothing named
Two alternatives
Pave, Salary.comFour alternatives
Culpepper, Pavilion GTM Compensation Benchmarks, Payscale, WorldatWorkSix alternatives
BCG, Bain & Company, Computer Economics, Gartner, KPMG, PwCThree alternatives
Pave, Ravio, SalaryExpertMistral SmallSalaryCubePavilion's B2B Tech Salary BenchmarksChangedDeloitteBureau of Labor Statistics (BLS) OEWSno first choiceagainst: Anthropic, OpenAI, UserBenchmark
Three alternatives
Pave, Payfactors, Salary.comFour alternatives
Compensation-IQ, Mercer, Ravio, Willis Towers WatsonSix alternatives
Bain & Company, Boston Consulting Group, Computer Economics, Gartner, KPMG, PwCThree alternatives
Comprehensive.io, Pave, SalaryExpertDeepSeek V4 FlashPayscaleagainst: Mercer, RadfordPayscale / CompAnalystChangedagainst: Mercer Benchmark Databaseno first choicePaveagainst: Glassdoor, Levels.fyi, Radford/Mercerno first choiceagainst: 3DMark, Cinebench 2024, Geekbench 6, MMLU, Open LLM Leaderboard, PCMark, PassMark PerformanceTest, UserBenchmark
Four alternatives
ERI, Pave, Ravio, Salary.comTwo alternatives
Radford, SalaryCubeFour alternatives
BLS Occupational Employment Statistics, Comprehensive.io, Payscale, RavioLlama 4 MaverickPayscaleno first choiceChangedno first choiceBLS OEWS, Comprehensive.io, Paveno first choicenothing named
Two alternatives
Ficstar, RavioQwen 3.7 FlashPayscaleagainst: Mercer, Radford, WTWMercer, Robert HalfChangedGartnerComprehensive.io, Paveagainst: Mercer, Towers Watsonno first choiceagainst: Deloitte, Glassdoor, Korn Ferry, Levels.fyi, LinkedIn, Mercer, WTWagainst: PassMark, UserBenchmark
Four alternatives
Pave, Payfactors, Salary.com, SalaryCubeFour alternatives
Payscale, Radford, Salary.com, Willis Towers WatsonThree alternatives
Computer Economics, Deloitte, ISGFive alternatives
BLS, Levels.fyi, Payscale, Ravio, Salary.comFour alternatives
CompAnalyst, Option Impact, Pave, RadfordKimi K2Payscale, SalaryCubePayscale / CompAnalystChangedagainst: Mercer, Radford, Willis Towers Watsonno first choiceComprehensive.io, U.S. Bureau of Labor Statisticsagainst: Betts, Kelly Services, Radford, Robert Halfno first choicenothing named
Three alternatives
Mercer, Radford, Salary.comTwo alternatives
CompData Surveys, Salary.com CompAnalystSeven alternatives
Computer Economics, Deloitte, Forrester, Gartner, IDC, KPMG, McKinsey & CompanyFour alternatives
Culpepper, Local/Regional HR Consulting Firms, Payscale, RavioGLM 4.7 FlashXPaveCompDataChangedno first choicePaveagainst: Glassdoor, Levels.fyi, Payscaleno first choicenothing named
Eight alternatives
Aon Radford, CaptivateIQ, Ficstar, Mercer, OpenComp, Payscale Ascent, Ravio, Salary.com CompAnalystTwo alternatives
Mercer, RadfordFive alternatives
BLS, Comprehensive.io, Ravio, Salary.com CompAnalyst, StartupCFO’s Startup Salary BenchmarkMiniMax M2.5Payscaleagainst: Mercer, RadfordCompAnalyst, PayscaleChangedno first choiceBLS Occupational Employment and Wage Statisticsno first choiceagainst: UserBenchmark
One alternative
Comprehensive.ioTwo alternatives
Mercer, RadfordFour alternatives
Figures, Glassdoor, Pave Market Data Lite, PayscaleBold is the first choiceAlternatives are counted; the count opens them.What the answer argued against
05The record
One row per call: the version string exactly as returned, whether the model searched, sources cited, and latency. Full answer text is in the free responses file. Download the record
Seventy-two rows: every prompt, every model, every answer.
| Prompt | Model | Version string | Time (UTC) | Searched | Sources | Latency |
|---|---|---|---|---|---|---|
| Direct recommendation | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-17 20:12 | yes | 9 | 8 s |
| Direct recommendation | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-17 17:59 | yes | 5 | 6 s |
| Direct recommendation | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-17 22:02 | yes | 13 | 53 s |
| Direct recommendation | Perplexity Sonar | sonar | 2026-09-17 20:41 | yes | 20 | 7 s |
| Direct recommendation | Grok 4.1 Fast | spacexai/grok-4.1-fast-non-reasoning via vertex | 2026-09-17 20:41 | yes | 34 | 19 s |
| Direct recommendation | Mistral Small | mistral/mistral-small via mistral | 2026-09-17 20:35 | yes | 5 | 4 s |
| Direct recommendation | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via fireworks | 2026-09-17 18:49 | yes | 15 | 22 s |
| Direct recommendation | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-17 21:33 | yes | 5 | 2 s |
| Direct recommendation | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-17 19:51 | yes | 5 | 25 s |
| Direct recommendation | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-17 19:21 | yes | 14 | 28 s |
| Direct recommendation | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-17 19:07 | yes | 11 | 64 s |
| Direct recommendation | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-17 19:42 | yes | 11 | 23 s |
| Paraphrase | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-17 21:13 | no | 0 | 4 s |
| Paraphrase | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-17 20:22 | no | 0 | 4 s |
| Paraphrase | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-17 18:50 | yes | 10 | 21 s |
| Paraphrase | Perplexity Sonar | sonar | 2026-09-17 21:36 | yes | 20 | 7 s |
| Paraphrase | Grok 4.1 Fast | spacexai/grok-4.1-fast-non-reasoning via vertex | 2026-09-17 18:45 | yes | 29 | 14 s |
| Paraphrase | Mistral Small | mistral/mistral-small via mistral | 2026-09-17 20:12 | yes | 5 | 5 s |
| Paraphrase | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via fireworks | 2026-09-17 21:42 | yes | 10 | 16 s |
| Paraphrase | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-17 21:27 | yes | 5 | 2 s |
| Paraphrase | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-17 19:12 | no | 0 | 29 s |
| Paraphrase | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-17 21:40 | yes | 19 | 29 s |
| Paraphrase | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-17 18:30 | yes | 23 | 107 s |
| Paraphrase | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-17 21:35 | yes | 5 | 27 s |
| Comparative | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-17 19:55 | yes | 9 | 8 s |
| Comparative | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-17 19:20 | no | 0 | 8 s |
| Comparative | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-17 18:33 | yes | 16 | 34 s |
| Comparative | Perplexity Sonar | sonar | 2026-09-17 21:41 | yes | 20 | 8 s |
| Comparative | Grok 4.1 Fast | spacexai/grok-4.1-fast-non-reasoning via vertex | 2026-09-17 20:25 | yes | 29 | 17 s |
| Comparative | Mistral Small | mistral/mistral-small via mistral | 2026-09-17 21:24 | yes | 5 | 7 s |
| Comparative | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via fireworks | 2026-09-17 20:50 | yes | 13 | 15 s |
| Comparative | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-17 19:59 | yes | 5 | 2 s |
| Comparative | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-17 18:55 | yes | 10 | 37 s |
| Comparative | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-17 21:46 | yes | 13 | 45 s |
| Comparative | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-17 21:04 | yes | 25 | 164 s |
| Comparative | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-17 21:08 | yes | 5 | 20 s |
| Budget constrained | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-17 19:03 | yes | 9 | 8 s |
| Budget constrained | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-17 18:05 | no | 0 | 3 s |
| Budget constrained | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-17 17:56 | yes | 14 | 23 s |
| Budget constrained | Perplexity Sonar | sonar | 2026-09-17 20:27 | yes | 20 | 4 s |
| Budget constrained | Grok 4.1 Fast | spacexai/grok-4.1-fast-non-reasoning via vertex | 2026-09-17 20:34 | yes | 32 | 15 s |
| Budget constrained | Mistral Small | mistral/mistral-small via mistral | 2026-09-17 21:18 | yes | 5 | 4 s |
| Budget constrained | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via fireworks | 2026-09-17 20:47 | yes | 9 | 13 s |
| Budget constrained | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-17 21:42 | yes | 7 | 4 s |
| Budget constrained | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-17 18:03 | yes | 5 | 17 s |
| Budget constrained | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-17 19:31 | yes | 9 | 34 s |
| Budget constrained | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-17 20:17 | yes | 21 | 100 s |
| Budget constrained | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-17 18:10 | yes | 5 | 15 s |
| Scale constrained | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-17 18:12 | no | 0 | 5 s |
| Scale constrained | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-17 21:30 | no | 0 | 9 s |
| Scale constrained | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-17 22:05 | yes | 12 | 58 s |
| Scale constrained | Perplexity Sonar | sonar | 2026-09-17 20:44 | yes | 20 | 7 s |
| Scale constrained | Grok 4.1 Fast | spacexai/grok-4.1-fast-non-reasoning via vertex | 2026-09-17 19:04 | yes | 13 | 10 s |
| Scale constrained | Mistral Small | mistral/mistral-small via mistral | 2026-09-17 20:33 | no | 0 | 9 s |
| Scale constrained | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via fireworks | 2026-09-17 19:07 | yes | 14 | 19 s |
| Scale constrained | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-17 20:35 | yes | 5 | 3 s |
| Scale constrained | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-17 21:33 | no | 0 | 29 s |
| Scale constrained | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-17 20:46 | yes | 14 | 30 s |
| Scale constrained | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-17 21:18 | yes | 14 | 50 s |
| Scale constrained | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-17 18:11 | yes | 10 | 45 s |
| Negative framing | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-17 17:58 | no | 0 | 2 s |
| Negative framing | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-17 18:10 | yes | 5 | 6 s |
| Negative framing | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-17 18:25 | yes | 22 | 23 s |
| Negative framing | Perplexity Sonar | sonar | 2026-09-17 21:20 | yes | 15 | 6 s |
| Negative framing | Grok 4.1 Fast | spacexai/grok-4.1-fast-non-reasoning via vertex | 2026-09-17 18:24 | yes | 39 | 18 s |
| Negative framing | Mistral Small | mistral/mistral-small via mistral | 2026-09-17 19:02 | yes | 5 | 6 s |
| Negative framing | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-09-17 19:30 | yes | 28 | 59 s |
| Negative framing | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-17 20:41 | yes | 5 | 3 s |
| Negative framing | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-17 21:07 | yes | 22 | 57 s |
| Negative framing | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-17 21:02 | yes | 15 | 35 s |
| Negative framing | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-17 19:40 | yes | 26 | 75 s |
| Negative framing | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-17 19:14 | yes | 15 | 38 s |
Normalization in this category
Every judgment call made between the raw labels and the numbers above, listed so it is visible and reversible.
Category-scoped readings
None. Every name in this category resolved on its own.
Unresolved, counted raw
AnandTech
Anthropic
BLS Occupational Employment and Wage Statistics (OEWS)
Betts
Cinebench 2024
CompTelecom
CrystalDiskMark
ERI (Economic Research Institute) Salary Assessor
ERI Economic Research Institute
Gamers Nexus
Geekbench 6
HELM
Kelly Services
LinkedIn job adverts
Local/Regional HR Consulting Firms
MMLU
Mercer Compensation Database
NERA
Open LLM Leaderboard
Oxera
PC-Kombo
PCMark (UL Solutions)
PassMark
PassMark PerformanceTest
Pavilion (formerly Revenue Collective)
Pavilion GTM Compensation Benchmarks
Pavilion's B2B Tech Salary Benchmarks
Pay Transparency Law Postings
Primate Labs
RepVue/Bridge Group-style data
S&P Global Mobility
SWE-bench Verified
StartupCFO’s Startup Salary Benchmark
Strategy&
TechPowerUp
TechSpot
Towers Watson
Trusted salary benchmarking surveys (imercer)
U.S. Bureau of Labor Statistics (BLS) - OEWS
Wage data from government sources
Willis Towers Watson [WTW]
WorldatWork
WorldatWork Research
Discontinued, still offered
No shut-down product was recommended here.