Data warehouse and reverse ETL for marketing
Asked as “marketing data warehouse and reverse ETL stack”, and as “reverse ETL and warehouse setup for a marketing team”, on behalf of a mid-market B2B company. 91 first choices recorded across the direct, paraphrase, budget and scale prompts, twelve models each.
Standing · first-choice share
42%
Clear leader
42Hightouch24BigQuery20Snowflake14others
40% of first choices across 43, the largest first-choice pool in the edition because answers here recommend stacks rather than single products. Only Hightouch clears the leader line.
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.
The 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
Show the eight products at 0%, ordered by negative rate
Bars are the share of first choices, 0 to 100Every product with at least 10 labels here. Every product name links to its product page.
BigQuery is made by Google, whose model Gemini 3.5 Flash is in the set. On the four share prompts that model made it the first choice three times of 4; the other eleven models nineteen times of 44. Lab treatment is defined on the method page; the row is marked, not excluded.
All fifteen head-to-head pages: the top six products, each against each
Hightouch vs RudderStack Reverse ETLHightouch vs PolytomicBigQuery vs SnowflakeBigQuery vs CensusBigQuery vs RudderStack Reverse ETLBigQuery vs PolytomicSnowflake vs CensusSnowflake vs RudderStack Reverse ETLSnowflake vs PolytomicCensus vs RudderStack Reverse ETLCensus vs PolytomicRudderStack Reverse ETL vs Polytomic
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
08
09
10
11
12
13
14Accepted challengerEndorsed leader
0%First-choice share → · lines at 30% share and 25% negative60%
Key
01Hightouch42%
Hightouch42%02BigQuery24%
BigQuery24%03Snowflake20%
Snowflake20%04Census5%
Census5%05RudderStack Reverse ETL1%
RudderStack Reverse ETL1%06Polytomic1%
Polytomic1%07Fivetran0%
Fivetran0%08Airbyte0%
Airbyte0%09dbt0%
dbt0%10Improvado0%
Improvado0%11Amazon Redshift0%
Amazon Redshift0%12Databricks SQL Warehouse0%
Databricks SQL Warehouse0%13Segment0%
Segment0%14Stitch0%
Stitch0%What they warned about
Five of twelve models held their first choice under the paraphrase. Claude Haiku 4.5, Gemini 3.5 Flash, Perplexity Sonar, Mistral Small, Llama 4 Maverick, Qwen 3.7 Flash and GLM 4.7 FlashX 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.
Amazon Redshift29%
10 of 35 labels negative · 9 of 12 models · 2 hard negative
“Avoid for tight budgets: Snowflake/Redshift (higher baseline compute ~$300+/month)” Grok 4.20, budget prompt
Census19%
12 of 63 labels negative · 8 of 12 models · 1 hard negative
“❌ Census: Requires Enterprise for most features; quote-based pricing” GLM 4.7 FlashX, budget prompt
Hightouch14%
9 of 65 labels negative · 8 of 12 models
“Both Census and Hightouch excel at the "reverse" direction but require separate solutions... This fragmentation increases complexity and maintenance burden.” Claude Haiku 4.5, negative prompt
BigQuery11%
7 of 65 labels negative · 7 of 12 models
“BigQuery (Proceed with Query Guardrails) ... can cost the company hundreds of dollars in a single click” Gemini 3.5 Flash, negative prompt
What they cite
Four of the twelve models returned a source list in this edition: four return the pages they cited.
Sites the answers cite
21 of 72 answers in this category came back with a source list, from 4 of 12 models: citations where the model returns them, or the search results it consulted. 321 links across 119 sites, every framing counted. Ranked by the number of answers carrying the site or page. The other 8 models searched through a gateway that returned no source list on this edition.
vendor site · Improvado15 answers · 36 citations · 3 models
vendor site · Hightouch11 answers · 21 citations · 4 models
vendor site · Cometly9 answers · 18 citations · 3 models
vendor site · Domo9 answers · 10 citations · 3 models
vendor site · Integrate7 answers · 11 citations · 2 models
vendor site · Supermetrics7 answers · 9 citations · 3 models
vendor site · LeadJourney7 answers · 7 citations · 3 models
vendor site · Funnel6 answers · 10 citations · 3 models
6 answers · 8 citations · 2 models
6 answers · 6 citations · 3 models
6 answers · 6 citations · 3 models
vendor site · Fastero5 answers · 5 citations · 3 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.
10 answers · 3 models
7 answers · 3 models
7 answers · 3 models
7 answers · 3 models
7 answers · 3 models
6 answers · 3 models
5 answers · 2 models
5 answers · 2 models
5 answers · 2 models
5 answers · 3 models
Search against answers
Each company's standing in the answers beside its site's footprint in Google search, one row a site: the products the models named on it with their shares, and the share they add up to; monthly searches on Google, and DataForSEO's estimate of AI search demand (modeled from search signals, directional, not a count of queries to any assistant), for the most-searched of the company's and its products' names (the name is in each row's hover text); estimated monthly organic visits to the site; and its best position in Google's top ten for “best marketing data warehouse and reverse etl stack”, “marketing data warehouse and reverse etl stack”. US estimates from DataForSEO and Google's Ads Transparency Center. A small company's site, or a mid-sized company's site for its flagship, is marked company; a product on a large parent's site (Google, Microsoft) has no site figures. A column with no figures for this category is left out, and an empty cell means none were seen, not none exist. Two measurements side by side: neither is read as the cause of the other.
| Company and products | Share | Own site cited | Name searches, Google | AI search demand, est. | Organic visits | Paid search a month, est. | Google ads, last 30 days | Best Google position |
|---|---|---|---|---|---|---|---|---|
| Hightouch hightouch.com | 42% | 399 | 12,100 | 434 | 127,772 | 23 | ||
| Google cloud.google.com BigQuery 24%, Looker 0% | 24% | 208 | 22,200 | 2,048 | 7,405,347 | |||
| Snowflake snowflake.com | 20% | 234 | 368,000 | 37,077 | 718,749 | 118 | ||
| Fivetran fivetran.com Census 5%, Fivetran 0% | 5% | 238 | 60,500 | 14,424 | 41,501 company | $126 | 97 | |
| Weld weld.app | 2% | 7 | 165,000 | 152,391 | 30,335 | 0 | ||
| RudderStack Reverse ETL rudderstack.com | 1% | 122 | 20 | 0 | 46,132 | 50 | 2 | |
| Polytomic polytomic.com | 1% | 59 | 590 | 16 | 882 | 1 | ||
| DuckDB duckdb.org | 1% | 12 | 18,100 | 686 | 15,468 | |||
| Adapters | 1% | |||||||
| ClicData clicdata.com | 1% |
Google's top ten for the two searches hold 17 sites; 4 of them are among the sites the models cited here (domo.com, getbruin.com, improvado.io, integrate.io). The first result for “best marketing data warehouse and reverse etl stack” is improvado.io.
How 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.ShowHide
How 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.
ShowHideModelDirect“What is the best marketing data warehouse and reverse ETL stack for a mid-market B2B company?”Paraphrase“Which reverse ETL and warehouse setup for a marketing team would you recommend to a mid-sized B2B company?”Comparative“What are the top data warehouse and reverse ETL tools for marketing teams and how do they differ?”Budget-constrained“What is the best marketing data warehouse and reverse ETL stack for a company with a limited budget?”Scale-constrained“We are a 500 person company evaluating a marketing data warehouse and reverse ETL stack. What should we look at?”Negative“Which data warehouse and reverse ETL tools for marketing teams should I avoid or be cautious about?”
Claude Haiku 4.5BigQuery, Hightouch, Polytomic, SnowflakeHightouchChangedBigQuery, HightouchBigQuery, ClicDatano first choiceagainst: Amazon Redshift, BigQuery, Census, Hightouch, Supermetrics
Three alternatives
Amazon Redshift, Hevo, RudderStack Reverse ETLFive alternatives
BigQuery, Census, Polytomic, RudderStack Reverse ETL, SnowflakeFive alternatives
Census, Fivetran, Microsoft Azure Synapse Analytics, Microsoft Fabric, SnowflakeThree alternatives
Hevo, Skyvia, SupermetricsGPT-5.4 miniHightouch, SnowflakeHightouch, SnowflakeHeldno first choiceBigQuery, HightouchCensus, Hightouch, Snowflakeagainst: Amazon Redshift, BigQuery, Snowflake
Five alternatives
BigQuery, Census, Databricks SQL Warehouse, Fivetran, dbtTwo alternatives
Amazon Redshift, BigQueryThree alternatives
Airbyte, PostgreSQL, RudderStack Reverse ETLTwo alternatives
BigQuery, Databricks SQL WarehouseGemini 3.5 FlashBigQuery, Hightouch, Snowflakeagainst: Salesforce Data Cloud, SegmentHightouch, SnowflakeChangedagainst: Segment, TealiumBigQuery, HightouchBigQuery, Hightouchagainst: Amazon Redshift, Census, Snowflake, dbtBigQuery, Hightouch, Snowflakeagainst: Databricks SQL Warehouse, Segment, Tealiumagainst: AWS Redshift, BigQuery, Census, Databricks SQL Warehouse, Hightouch, Multywoven, RudderStack Reverse ETL, Snowflake, Stitch
Three alternatives
Census, Fivetran, dbtFive alternatives
Airbyte, BigQuery, Census, Fivetran, dbtFour alternatives
Census, Databricks SQL Warehouse, RudderStack Reverse ETL, SnowflakeThree alternatives
Airbyte, Fivetran, dbtFour alternatives
Airbyte, Census, Fivetran, dbtPerplexity SonarBigQuery, HightouchHightouchChangedBigQuery, Census, Hightouch, SnowflakeBigQuery, Weldagainst: Supermetrics, Whatagraphno first choiceagainst: Airbyte, Amazon Redshift, BigQuery, Census, Fivetran, Hightouch, Snowflake, Stitch
Six alternatives
Cometly, Fivetran, Funnel, Improvado, LeadJourney, SegmentFour alternatives
BigQuery, Census, Snowflake, dbtThree alternatives
Amazon Redshift, Azure Synapse Analytics, RudderStack Reverse ETLFour alternatives
Airbyte, Metabase, PostgreSQL, dbtGrok 4.20BigQuery, HightouchBigQuery, HightouchHeldBigQuery, HightouchHightouchagainst: Amazon Redshift, SnowflakeBigQuery, Census, Hightouch, Snowflakeagainst: Amazon Redshift, BigQuery, Census, Databricks SQL Warehouse, Fivetran, Hightouch, RudderStack Reverse ETL, Segment, Snowflake
Two alternatives
Census, SnowflakeSix alternatives
Airbyte, Census, Fivetran, Looker Studio, Snowflake, dbtFour alternatives
Amazon Redshift, Census, RudderStack Reverse ETL, SnowflakeFive alternatives
Airbyte, BigQuery, Census, Grouparoo, Supabase PostgresFour alternatives
Airbyte, Fivetran, Looker, dbtMistral SmallBigQuery, HightouchHightouch, SnowflakeChangedagainst: Amazon RedshiftHightouch, SnowflakeDuckDB, Multiwoven, RudderStack Reverse ETLBigQuery, Hightouchagainst: Funnel, Hevo, Improvado, Informatica, Polytomic, Talend
Seven alternatives
Airbyte, Census, Databricks SQL Warehouse, Fivetran, Improvado, Snowflake, dbtFour alternatives
Airbyte, BigQuery, Census, FivetranEight alternatives
Amazon Redshift, BigQuery, Census, Databricks SQL Warehouse, Fivetran, Improvado, LeadJourney, dbtFive alternatives
Amazon Redshift, BigQuery, Census, ClickHouse, PolytomicSix alternatives
Census, Fivetran, Segment, Snowflake, dbt, mParticleDeepSeek V4 FlashBigQuery, HightouchBigQuery, HightouchHeldBigQuery, HightouchAdapters, BigQueryagainst: Census, Fivetran, Hightouch, Snowflake, Stitch, TableauHightouch, Snowflakeagainst: Databricks SQL Warehouse, Stitchagainst: Amazon Redshift, Azure Synapse Analytics, Census, Fivetran, Grouparoo, Hightouch, IBM Db2 Warehouse, Oracle ADW, RudderStack Reverse ETL, SAP Datasphere, Teradata
Six alternatives
Airbyte, Census, Fivetran, PostgreSQL, Snowflake, dbtNine alternatives
Airbyte, Amazon Redshift, Census, Fivetran, Looker, Metabase, PostgreSQL, Snowflake, dbtFive alternatives
Amazon Redshift, Census, Polytomic, RudderStack Reverse ETL, SnowflakeEight alternatives
Airbyte, Dagster, Dataddo, Looker Studio, Metabase, PostgreSQL, dbt, dltSix alternatives
Airbyte, BigQuery, Census, Fivetran, RudderStack Reverse ETL, dbtLlama 4 Maverickno first choiceCensus, HightouchChangedHightouchno first choiceno first choicenothing named
Two alternatives
BigQuery, SnowflakeFour alternatives
Census, Fivetran, Polytomic, RudderStack Reverse ETLQwen 3.7 FlashHightouch, SnowflakeBigQuery, HightouchChangedagainst: Oracle Autonomous Data Warehouse, Teradatano first choiceWeldagainst: Hightouch, MeltanoCensus, Hightouchagainst: Salesforce Data Cloudagainst: BigQuery, Databricks SQL Warehouse, Teradata
Six alternatives
BigQuery, Census, Fivetran, Improvado, Polytomic, dbtFour alternatives
Airbyte, Fivetran, Snowflake, dbtFive alternatives
BigQuery, Fivetran, Jitsu, Polytomic, SnowflakeFour alternatives
BigQuery, Databricks SQL Warehouse, Snowflake, dbtKimi K2Hightouch, Snowflakeagainst: Databricks SQL WarehouseHightouch, SnowflakeHeldagainst: CensusBigQuery, Census, Hightouch, SnowflakeBigQuery, Hightouchagainst: Census, dbtHightouch, Snowflakeagainst: Segmentagainst: Amazon Redshift, BigQuery, Census, Hex, Hightouch, Looker Studio, Microsoft Power BI, Mode, Polytomic
Eight alternatives
Airbyte, BigQuery, Census, Fivetran, Hevo, Improvado, Polytomic, dbtThree alternatives
BigQuery, Fivetran, dbtSeven alternatives
Cometly, DinMo, Fivetran, Funnel, Polytomic, RudderStack Reverse ETL, SkyviaFive alternatives
Airbyte, Definite, Metabase, Redshift Serverless, dbtFive alternatives
Airbyte, BigQuery, Census, Fivetran, RudderStack Reverse ETLGLM 4.7 FlashXHightouch, SnowflakeBigQuery, HightouchChangedHightouch, SnowflakeBigQuery, Hightouchagainst: Census, Fivetran, Polytomic, SnowflakeBigQuery, Hightouchagainst: Excel/Spreadsheets, Looker Studio, Microsoft Power BI, Tableau
Five alternatives
BigQuery, Census, Databricks SQL Warehouse, Polytomic, dbtFour alternatives
Census, Improvado, Segment Reverse ETL, SnowflakeSix alternatives
Amazon Redshift, BigQuery, Census, Domo, Fivetran, ImprovadoTwo alternatives
Airbyte, RudderStack Reverse ETLFive alternatives
Amazon Redshift, Census, Fivetran, Snowflake, WeldMiniMax M2.5Hightouch, Snowflakeagainst: Amazon Redshift, SegmentHightouch, SnowflakeHeldHightouchBigQuery, Census, Hightouch, SnowflakeHightouchagainst: Census, Hightouch
Two alternatives
BigQuery, CensusSix alternatives
Amazon Redshift, BigQuery, Census, Fivetran, GrowthLoop, HevoFour alternatives
Census, Domo, Fivetran, RudderStack Reverse ETLTwo alternatives
Airbyte, StitchFive alternatives
BigQuery, Census, Fivetran, Snowflake, dbtBold is the first choiceAlternatives are counted; the count opens them.What the answer argued against
The record
One row per call: the version string exactly as returned, whether the model searched, sources cited, and latency.
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 | 2026-09-15 04:50 | yes | 15 | 10 s |
| Direct recommendation | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-15 04:39 | yes | 7 | 6 s |
| Direct recommendation | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-15 00:13 | yes | 12 | 23 s |
| Direct recommendation | Perplexity Sonar | sonar | 2026-09-15 01:39 | yes | 20 | 8 s |
| Direct recommendation | Grok 4.20 | spacexai/grok-4.1-fast-non-reasoning via vertex (asked for grok-4.20-non-reasoning; the API answered as this version) | 2026-09-15 00:17 | yes | 0 | 10 s |
| Direct recommendation | Mistral Small | mistral/mistral-small via mistral | 2026-09-15 00:47 | yes | 0 | 10 s |
| Direct recommendation | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via fireworks | 2026-09-15 04:47 | yes | 0 | 23 s |
| Direct recommendation | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-15 03:57 | yes | 0 | 2 s |
| Direct recommendation | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-15 04:02 | yes | 0 | 53 s |
| Direct recommendation | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-15 03:15 | yes | 0 | 44 s |
| Direct recommendation | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-15 01:02 | yes | 0 | 127 s |
| Direct recommendation | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-15 03:36 | yes | 0 | 26 s |
| Paraphrase | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-15 00:13 | yes | 20 | 11 s |
| Paraphrase | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-15 01:12 | yes | 5 | 6 s |
| Paraphrase | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-15 00:37 | yes | 19 | 30 s |
| Paraphrase | Perplexity Sonar | sonar | 2026-09-15 00:03 | yes | 20 | 11 s |
| Paraphrase | Grok 4.20 | spacexai/grok-4.1-fast-non-reasoning via vertex (asked for grok-4.20-non-reasoning; the API answered as this version) | 2026-09-15 03:21 | yes | 0 | 8 s |
| Paraphrase | Mistral Small | mistral/mistral-small via mistral | 2026-09-15 04:16 | yes | 0 | 10 s |
| Paraphrase | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via fireworks | 2026-09-15 03:36 | yes | 0 | 20 s |
| Paraphrase | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-15 00:56 | yes | 0 | 3 s |
| Paraphrase | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-14 23:15 | yes | 0 | 32 s |
| Paraphrase | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-14 23:10 | yes | 0 | 46 s |
| Paraphrase | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-15 01:25 | yes | 0 | 77 s |
| Paraphrase | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-15 02:54 | yes | 0 | 12 s |
| Comparative | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-15 00:26 | yes | 22 | 13 s |
| Comparative | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-15 00:45 | yes | 6 | 7 s |
| Comparative | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-15 03:13 | yes | 13 | 21 s |
| Comparative | Perplexity Sonar | sonar | 2026-09-15 03:56 | yes | 20 | 9 s |
| Comparative | Grok 4.20 | spacexai/grok-4.1-fast-non-reasoning via vertex (asked for grok-4.20-non-reasoning; the API answered as this version) | 2026-09-15 04:17 | yes | 0 | 8 s |
| Comparative | Mistral Small | mistral/mistral-small via mistral | 2026-09-15 00:25 | yes | 0 | 11 s |
| Comparative | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-09-15 02:55 | yes | 0 | 53 s |
| Comparative | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-15 04:58 | yes | 0 | 2 s |
| Comparative | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-15 03:21 | yes | 0 | 20 s |
| Comparative | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-15 04:52 | yes | 0 | 45 s |
| Comparative | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-15 05:04 | yes | 0 | 29 s |
| Comparative | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-15 01:28 | yes | 0 | 11 s |
| Budget constrained | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-15 02:24 | yes | 13 | 11 s |
| Budget constrained | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-15 04:58 | yes | 5 | 6 s |
| Budget constrained | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-14 23:01 | yes | 14 | 30 s |
| Budget constrained | Perplexity Sonar | sonar | 2026-09-15 03:39 | yes | 20 | 7 s |
| Budget constrained | Grok 4.20 | spacexai/grok-4.1-fast-non-reasoning via vertex (asked for grok-4.20-non-reasoning; the API answered as this version) | 2026-09-14 23:41 | yes | 0 | 10 s |
| Budget constrained | Mistral Small | mistral/mistral-small via mistral | 2026-09-14 22:56 | yes | 0 | 12 s |
| Budget constrained | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via fireworks | 2026-09-15 04:37 | yes | 0 | 22 s |
| Budget constrained | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-15 02:56 | yes | 0 | 1 s |
| Budget constrained | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-15 01:20 | yes | 0 | 51 s |
| Budget constrained | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-15 02:29 | yes | 0 | 42 s |
| Budget constrained | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-15 04:24 | yes | 0 | 139 s |
| Budget constrained | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-15 01:04 | yes | 0 | 15 s |
| Scale constrained | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-15 00:24 | no | 0 | 9 s |
| Scale constrained | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-15 02:46 | yes | 3 | 9 s |
| Scale constrained | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-15 03:04 | no | 0 | 16 s |
| Scale constrained | Perplexity Sonar | sonar | 2026-09-15 02:38 | yes | 20 | 9 s |
| Scale constrained | Grok 4.20 | spacexai/grok-4.1-fast-non-reasoning via vertex (asked for grok-4.20-non-reasoning; the API answered as this version) | 2026-09-15 00:04 | yes | 0 | 11 s |
| Scale constrained | Mistral Small | mistral/mistral-small via mistral | 2026-09-15 03:24 | yes | 0 | 14 s |
| Scale constrained | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-09-15 01:01 | yes | 0 | 51 s |
| Scale constrained | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-15 04:01 | yes | 0 | 2 s |
| Scale constrained | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-15 04:22 | yes | 0 | 40 s |
| Scale constrained | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-15 01:12 | yes | 0 | 63 s |
| Scale constrained | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-15 02:24 | yes | 0 | 18 s |
| Scale constrained | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-15 00:46 | yes | 0 | 14 s |
| Negative framing | Claude Haiku 4.5 | claude-haiku-4-5-20251001 | 2026-09-14 23:49 | yes | 32 | 14 s |
| Negative framing | GPT-5.4 mini | gpt-5.4-mini-2026-03-17 | 2026-09-15 02:37 | yes | 0 | 7 s |
| Negative framing | Gemini 3.5 Flash | gemini-3.5-flash | 2026-09-15 02:34 | yes | 15 | 24 s |
| Negative framing | Perplexity Sonar | sonar | 2026-09-15 03:17 | yes | 20 | 8 s |
| Negative framing | Grok 4.20 | spacexai/grok-4.1-fast-non-reasoning via vertex (asked for grok-4.20-non-reasoning; the API answered as this version) | 2026-09-15 04:02 | yes | 0 | 10 s |
| Negative framing | Mistral Small | mistral/mistral-small via mistral | 2026-09-15 03:33 | yes | 0 | 11 s |
| Negative framing | DeepSeek V4 Flash | deepseek/deepseek-v4-flash via deepinfra | 2026-09-15 01:08 | yes | 0 | 120 s |
| Negative framing | Llama 4 Maverick | meta/llama-4-maverick via bedrock | 2026-09-15 04:24 | yes | 0 | 3 s |
| Negative framing | Qwen 3.7 Flash | alibaba/qwen3.7-flash via alibaba | 2026-09-15 01:20 | no | 0 | 47 s |
| Negative framing | Kimi K2 | moonshotai/kimi-k2 via novita | 2026-09-14 22:58 | yes | 0 | 40 s |
| Negative framing | GLM 4.7 FlashX | zai/glm-4.7-flashx via zai | 2026-09-15 04:30 | yes | 0 | 28 s |
| Negative framing | MiniMax M2.5 | minimax/minimax-m2.5 via minimax | 2026-09-15 04:22 | yes | 0 | 42 s |
Noise floor in this category
Flips between the edition run and its calibration repeat. Six prompts per model is a small sample; the index-wide floor is the number to trust.
ShowHide
Claude Haiku 4.5
6 of 6 flipped
GPT-5.4 mini
3 of 5 flipped
Gemini 3.5 Flash
4 of 5 flipped
Perplexity Sonar
2 of 4 flipped
Grok 4.20
4 of 6 flipped
Mistral Small
6 of 6 flipped
DeepSeek V4 Flash
3 of 6 flipped
Llama 4 Maverick
3 of 3 flipped
Qwen 3.7 Flash
5 of 6 flipped
Kimi K2
3 of 5 flipped
GLM 4.7 FlashX
5 of 6 flipped
MiniMax M2.5
4 of 6 flipped
Normalization in this category
Every judgment call made between the raw labels and the numbers above, listed so it is visible and reversible.
ShowHide
Category-scoped readings
Amazon Redshift Serverless read as Amazon Redshift
Azure Synapse read as Microsoft Azure Synapse Analytics
Databricks read as Databricks SQL Warehouse
Databricks (Lakehouse) read as Databricks SQL Warehouse
Microsoft Azure Synapse read as Microsoft Azure Synapse Analytics
Oracle read as Oracle Autonomous Data Warehouse
RudderStack read as RudderStack Reverse ETL
Segment's reverse ETL read as Segment Reverse ETL
Unresolved, counted raw
Adapters
Excel/Spreadsheets
Multywoven
Neon/Supabase
Oracle ADW
Raygun
SAP
Sheets/Looker Studio
Supabase Postgres
Twilio Segment Reverse ETL / Profiles Sync
Discontinued, still offered
Grok 4.20 named Grouparoo as alternative on the budget prompt. Open source reverse ETL, shut down 2023.
MiniMax M2.5 named Grouparoo as alternative on the negative prompt. Open source reverse ETL, shut down 2023.