Four of twelve models named Intercom first on the direct prompt; one named Customer.io. Intercom was named by twelve of the twelve models and Customer.io by ten and Intercom carries 38 labels and Customer.io 16, so the shares are not directly comparable.
United States, founded 2011. Named in ten categories this edition.
Named in seven categories this edition.
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; rank is within the category; every quote names the model and the prompt it came from. Both figures come from the in-app messaging and customer communications page.
| Model | Direct | Paraphrase | Comparative | Budget-constrained | Scale-constrained | Negative |
|---|---|---|---|---|---|---|
| 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 | ||||||
| MiniMax M2.5 |
Bold names in an answer are the products the judge labeled a first choice; a model naming several gives each of them that label. The full answer text for every row is in the record.
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. Two of two in this category shown.
“If you are a small-to-medium business on a tight or predictable budget, avoid Intercom's default AI tiers.” Gemini 3.5 Flash · negative prompt · hard negative
“I'd usually recommend Intercom if your goal is product messaging plus broader customer communications in one place.” GPT-5.4 mini · paraphrase prompt · first choice
Every negative label with a quote, up to three, then the highest-weighted positives, up to three. One of one in this category shown.
“Can be overkill compared with PLG-focused tools” Perplexity Sonar · negative prompt · soft negative
Comparisons are drawn for the top eight products in each category, each against each. The output is the models' output; nothing here is a recommendation by the index.