AI assistants that let you choose the AI model
Which AI assistants let me pick the underlying model?
Updated · 3 assistants ranked
Three assistants in our catalog say you can pick the model behind them: Luke (a choice of 200+ models), OpenClaw (any model, hosted or local) and OpenInstinct (any model through Vercel AI Gateway). All three took at least four of our ten tests: Luke 7.2 over 6, OpenInstinct 6.0 over 5, OpenClaw 5.1 over 10.
Top picks
- 17.2
Tested on 6 of 10Best: Travel9, Phone calls8Time to finish 9m 6s
- 26.0OpenInstinctBest free option
Tested on 5 of 10Best: Online tasks10, Permissions7
- 35.1
Tested on 10 of 10Best: Online tasks7, Routines7
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 | Luke | 6 | 9 | — | 8 | 7 | — | 7 | — | — | 6 | 7.2 |
| 2 | OpenInstinct | 10 | — | — | — | 3 | — | — | 6 | 7 | 4 | 6.0 |
| 3 | 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) says you choose the model; assistants that list several providers without saying you choose, such as Neo, Airtap and Friday, are left out. They are ranked by the average of the tests they took. Scores come from the published score sheet and are dated 2026-10-27. 2 of the 3 ranked assistants (Luke and OpenClaw) carry pre-filled scores, which are placeholder estimates the site has not yet replaced with its own test runs, so treat their places as provisional. OpenInstinct is not pre-filled.
How to choose
- Remember that the model is yours to pay for. OpenClaw says you pay your own model provider and OpenInstinct that you pay the underlying services; Luke is listed as free with no allowance stated.
- Treat the score as a guide to the assistant, not to every model it can run: the model you pick is not recorded in our scores.
- Check where it runs. OpenClaw and OpenInstinct are open source; Luke is a voice-driven Mac assistant that reaches WhatsApp and Telegram.
Which AI assistants let you choose the model?
Luke (200+ models), OpenClaw (any model, hosted or local) and OpenInstinct (any model available through Vercel AI Gateway), based on each catalog entry as checked on 2026-10-26.
Does the model decide the score?
Not alone. Luke averages 7.2 over six tests, OpenInstinct 6.0 over five and OpenClaw 5.1 over ten, with different model setups behind each.
Can I run one locally?
OpenClaw names local providers among the models you can use; see the report on assistants that run locally.
What users say
the memory just… drifts. Had it handling a simple multi-step task and it straight-up forgot a key detail from the same conversation 20 minutes earlier.
Open sourceIt figured out the Spotify API, wrote the skill itself, and now it just pings me.
Open sourceI just setup and it ordered me lunch, running on my own infra.
Open sourceFrequently asked questions
Which assistant is best for this?
Luke leads with an average of 7.2, ahead of OpenInstinct and OpenClaw.
Is there a good free option?
Luke is the highest-ranked assistant with a free plan, averaging 7.2.
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.