AI assistants that list more than one AI model
Which AI assistants say they use several AI models or providers, and how do they score across our ten tests?
8 assistants ranked
Eight assistants in our catalog list more than one model or provider and took at least four of our ten tests. Neo (8.1 over all ten tests) leads, then Ollie (7.8 over eight tests) and Luke (7.2 over six tests). Airtap (6.5 over four tests), Buoy (6.3 over eight tests), OpenInstinct (6.0 over five tests), Friday (5.8 over six tests) and OpenClaw (5.1 over all ten tests) follow.
Top picks
- 18.1NeoTop pick
Tested on 10 of 10Best: Travel9, Phone calls9
- 27.8
Tested on 8 of 10Best: Online tasks9, Routines9Time to finish 7m 25s
- 37.2
Tested on 6 of 10Best: Travel9, Phone calls8
- 46.5
Tested on 4 of 10Best: Routines8, Travel7
- 56.3
Tested on 8 of 10Best: Purchasing8, Online tasks7
Top pick: highest average here. Best free option: highest among assistants you can use for free. Fastest that's still good: quickest median time to finish among those averaging 7 or more.
Full ranking
| # | Assistant | Online tasks | Travel | Purchasing | Phone calls | Proactive | Routines | Connected apps | Permissions | Memory | Average | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Neo | 7 | 9 | 8 | 9 | 6 | 9 | 8 | 9 | 8 | 8 | 8.1 |
| 2 | Ollie | 9 | 7 | 7 | N/A | 7 | 8 | 9 | 7 | 8 | — | 7.8 |
| 3 | Luke | 6 | 9 | — | 8 | 7 | — | 7 | — | — | 6 | 7.2 |
| 4 | Airtap | — | 7 | — | — | — | 6 | 8 | 5 | — | — | 6.5 |
| 5 | Buoy | 7 | 6 | 8 | — | — | 4 | 7 | 6 | 7 | 5 | 6.3 |
| 6 | OpenInstinct | 10 | — | — | — | 3 | — | — | 6 | 7 | 4 | 6.0 |
| 7 | Friday | — | 5 | — | 5 | — | — | 6 | 7 | 6 | 6 | 5.8 |
| 8 | OpenClaw | 7 | 3 | 6 | 3 | 4 | 5 | 7 | 7 | 4 | 5 | 5.1 |
How this report works
Included are assistants whose catalog model line (checked 2026-10-26) names two or more providers or models, or lets you pick among them, with at least four of our ten tests taken, ranked by the average of the tests they took. The model line comes from each product's own pages and does not tell you which model answers a given request.
How to choose
- Separate who picks the model. Luke (200+ models), OpenInstinct (any model through Vercel AI Gateway) and OpenClaw (hosted or local models) let you choose; Neo, Ollie, Airtap, Friday and Buoy list several providers but choose for you.
- Read the provider lists. Neo names Claude, GPT and Gemini; Ollie Anthropic, Google and OpenAI; Airtap Google, Anthropic, xAI, OpenAI, Groq and AssemblyAI; Friday Anthropic, Gemini and Azure OpenAI; Buoy lists Anthropic, OpenAI, Google Cloud AI and Perplexity without naming models.
- Read the tests-taken count. Neo and OpenClaw took all ten tests, Ollie and Buoy eight, Friday six and Airtap only four.
Which AI assistants use more than one model?
Eight in our catalog list several models or providers. Neo 8.1, Ollie 7.8 and Luke 7.2 lead on average.
Can I choose the model myself?
Only Luke, OpenInstinct and OpenClaw say so. The others list several providers but we do not know which one answers a given request.
Does using several models make an assistant better?
Not on our data. Neo 8.1 leads, but OpenClaw, which lets you pick any model, averages 5.1 over all ten tests.
What users say
From there Buoy searched X, found my linked GitHub page (which currently 404s as I privated it), and even decided to look at my HuggingFace profile somehow. I’ve never seen an AI agent so willing to be proactive.
Open sourceBuoy says passwords stay out of chat and logs. Its privacy notice lists passwords among data it may collect. Not proof of exposure.
Open sourceUse Ollie if you want a text-only assistant that quietly watches your inbox and calendar and pings you; use Ohai if you want an app-based shared household workspace the whole family uses together.
Open sourceFrequently asked questions
Which assistant is best for this?
Neo leads with an average of 8.1, ahead of Ollie and Luke.
Is there a good free option?
Neo is the highest-ranked assistant with a free plan, averaging 8.1.
How did you test these assistants?
Every assistant got the same published tests, word for word, scored from three checkpoints and the number of corrections it needed. Each test is on the Dimensions page.