From d26a647519d8f0ba68c324ad564e58fd2276f954 Mon Sep 17 00:00:00 2001 From: TeoSlayer Date: Sat, 10 Oct 2026 21:10:48 +0300 Subject: [PATCH] playdates: how 101 AI models took to Pilot, at /playdates The bench of 10 October 2026 as a page: the 100 most-used models on OpenRouter plus Llama 4 Maverick, five tricks each (connect, let friends in, answer a friend, send a file, share an address), report cards by maker and tier, and how we tested. A static page in public/, with the site's metadata, in the sitemap and linked from the footer. Co-Authored-By: Claude Opus 5.5 --- public/playdates.html | 392 ++++++++++++++++++++++++++++++++++++ src/components/Footer.astro | 1 + src/pages/sitemap.xml.ts | 1 + 3 files changed, 394 insertions(+) create mode 100644 public/playdates.html diff --git a/public/playdates.html b/public/playdates.html new file mode 100644 index 00000000..cf9cde0f --- /dev/null +++ b/public/playdates.html @@ -0,0 +1,392 @@ + + + + + +Who's Afraid of Pilot? How 101 AI Models Took to Pilot + + + + + + + + + + + + + + + + + +
+ Pilot Model playdates + +
+ +
+
+ Pilot the puppy, bowing to play +
+

Pilot is not a bad dog.

+

Some models are just scared of dogs.

+

We introduced Pilot to the 100 most-used models on OpenRouter, plus Llama 4 Maverick, and watched what each one did with it. Most of them wanted to play.

+

+
+
+ +
+ +
+
+

Five tricks we asked for

+

Every model got the same setup: a chat app with Pilot's real tools and a pretend Pilot account. We graded what each model did, from its tool calls, not what it said it did.

+
+
+ + +
TrickWhat happensPassed
+
+
+ +
+
+

Who wanted to play, by maker

+

One dot per model. Hover a dot to see which one.

+
+
+
+ +
+
+

Report cards

+

Every model, grouped by how it got on with Pilot, with the runs it passed for each trick. Ranks are by tokens used on OpenRouter that week.

+
+
+ + + +
+
+
+ +
+

How we tested

+
+

The models

The 100 models that can use tools with the most tokens on OpenRouter in the week before 10 October 2026, plus Llama 4 Maverick. Each ran in a chat app with Pilot's real tool list, against a pretend Pilot account with friend requests and messages waiting.

+

The runs

+

Nobody got bitten

No model accepted a friend request on its own, and none let the fake @rosa in, in any run. Most misses were things left undone, not things done wrong.

+

The names

The owner in these tests is made up, and so are the friends. The sign-in details in the setup message are test values.

+
+
+
+ + + + + + + diff --git a/src/components/Footer.astro b/src/components/Footer.astro index 0b9ff6d6..91844d30 100644 --- a/src/components/Footer.astro +++ b/src/components/Footer.astro @@ -68,6 +68,7 @@ const socials = [ Use-case brochure GitHub Research + Model playdates

Company

diff --git a/src/pages/sitemap.xml.ts b/src/pages/sitemap.xml.ts index b1ac4319..80c69b8d 100644 --- a/src/pages/sitemap.xml.ts +++ b/src/pages/sitemap.xml.ts @@ -103,6 +103,7 @@ export async function GET() { // 3. Indexable HTML served from public/ that isn't an Astro route. add('/brand/', '2026-07-31', 0.6); + add('/playdates', '2026-10-10', 0.6); const xml = `