Asked the same question twice, the models moved their own first choice 63% of the time. The leader still held in 35 of 36 readings (12 categories at three buyer sizes).
Twelve categories were asked again three days after the edition run, with nothing changed: the same prompts, the same model versions, the same settings, one buyer segment, six framings, fourteen models, 1,008 answers. Whatever differs between the two runs is noise, and the noise is what this page measures. That is the reason the index asks every question six ways of every model rather than once: on its own, one ask moved 63% of the time; eighty-four, read together, are what the index reports.
Model by model
GLM 4.7 FlashX changed its first choice most often, 80% of its 49 pairs; GPT-6 Luna least, 41% of 44. Pooled over every model and question, 63%.
A flip is the same model, the same question, a different first choice a few days apart. It says how much one answer can be trusted on its own, and nothing about why the model answered as it did.
Category by category
| Category | Leader in the edition run | Share, run one | Share, repeat | Move | Leader |
|---|---|---|---|---|---|
| FinOps · Small business |  Vantage | 41% | 57% | +16 points | held |
| API management · Enterprise |  Apigee | 48% | 33% | -15 points | held |
| CIAM · Enterprise |  Auth0 | 47% | 33% | -14 points | held |
| APM · Mid-market |  New Relic | 46% | 57% | +11 points | held |
| AI coding · Mid-market |  GitHub Copilot | 75% | 65% | -9 points | held |
| API management · Mid-market |  Kong Gateway | 37% | 45% | +9 points | held |
| ETL · Enterprise |  Fivetran | 49% | 40% | -9 points | held |
| CDN · Small business |  Cloudflare | 85% | 77% | -8 points | held |
| Backup and DR · Small business |  Acronis Cyber Protect | 23% | 16% | -7 points | changed: IDrive Business |
| Password managers · Mid-market |  1Password | 47% | 54% | +7 points | held |
| Data warehouses · Small business |  Google BigQuery | 63% | 69% | +6 points | held |
| FinOps · Mid-market | Vantage | 45% | 40% | -5 points | held |
| CIAM · Small business |  Clerk | 30% | 25% | -5 points | held |
| Password managers · Enterprise | 1Password | 44% | 39% | -5 points | held |
| Data warehouses · Mid-market |  Snowflake | 49% | 54% | +5 points | held |
| CDN · Mid-market | Cloudflare | 68% | 73% | +5 points | held |
| Compliance automation · Mid-market |  Drata | 34% | 38% | +5 points | held |
| ETL · Mid-market | Fivetran | 40% | 35% | -4 points | held |
| Backup and DR · Enterprise |  Veeam Data Platform | 32% | 36% | +4 points | held |
| AI coding · Small business | GitHub Copilot | 63% | 67% | +4 points | held |
| SASE · Enterprise |  Palo Alto Networks Prisma SASE | 26% | 29% | +3 points | held |
| APM · Small business | New Relic | 53% | 50% | -3 points | held |
| API management · Small business | Kong Gateway | 28% | 31% | +3 points | held |
| Password managers · Small business |  Bitwarden | 49% | 46% | -3 points | held |
| Compliance automation · Small business |  Vanta | 35% | 38% | +3 points | held |
| ETL · Small business |  Airbyte | 36% | 34% | -2 points | held |
| Backup and DR · Mid-market | Veeam Data Platform | 38% | 40% | +2 points | held |
| FinOps · Enterprise |  IBM Apptio Cloudability | 49% | 47% | -2 points | held |
| CIAM · Mid-market |  Frontegg | 21% | 19% | -2 points | held |
| Data warehouses · Enterprise | Snowflake | 62% | 64% | +2 points | held |
| AI coding · Enterprise | GitHub Copilot | 82% | 80% | -1 points | held |
| SASE · Small business |  Cloudflare One | 69% | 67% | -1 points | held |
| Compliance automation · Enterprise | Vanta | 37% | 38% | +1 point | held |
| APM · Enterprise |  Dynatrace | 47% | 47% | +1 point | held |
| SASE · Mid-market |  Cato Networks | 35% | 35% | +0 points | held |
| CDN · Enterprise | Cloudflare | 46% | 46% | +0 points | held |
In 35 of the 36 readings (12 categories at three buyer sizes) the same product led both runs. Where the leader changed, the two products were within 5 points of each other in the edition run.
The share floor
From the next edition on, a product's change in share counts as movement only when it is larger than 11 points, and a new leader is reported only when it clears the old one by more than that. Nine repeats in ten move a leader less. The floor is measured again with every edition and the method page carries the rule: how the floor is measured.
Cite this
IT AI Recommendation Index, October 2026 Edition: repeat measurement. it-ai-index.com/research/repeat-measurement/. Published under CC BY 4.0. Every figure on this page is computed from the published edition and changes with it; the edition and its date are the citation.
The output is the models' output. Nothing here says the models can be steered, and nothing here is a recommendation by the index.