MODELDRIFT.WATCH

Sep 22 · Day 13

MINOR · CLAUDE · QUESTION 31

Claude named a different first choice for a question about which JavaScript framework to start with.

Why it’s minor

Which products a model recommends is read out of its answer by the least reliable part of our scorer, so compare the two answers before reading much into this.

The full answers

Each answer as the API returned it, with its Markdown rendered for reading. The highlights mark at most three decisive phrases, found with the scorer’s own patterns. Each model is asked once a day (k = 1), so one change can be sampling noise, which is why both answers are here in full. How scoring works.

Asked verbatim · Question 31

Which JavaScript framework should I start a new project with today?

SEP 21beforePermalink

Picks read from the answer, in order
  1. Astro
  2. Svelte + SvelteKit
  3. Angular
  4. Middle ground between React and Svelte

Short answer: React with a meta-framework is the safest default, but it's genuinely not the best pick for every project. Here's how I'd actually decide.

Default choice: React + a meta-framework
Largest ecosystem, most third-party components, easiest hiring, most Stack Overflow/LLM training data when you get stuck. For the meta-framework: Next.js if you want the most-traveled path, React Router (v7, formerly Remix) if you want something simpler and less coupled to a hosting vendor, or TanStack Start if you like that ecosystem. The main cost is complexity — React Server Components and the client/server boundary confuse experienced developers, not just beginners.

Pick something else if:

  • •Content-heavy site (blog, docs, marketing) → Astro. Ships almost no JS by default and lets you drop in React/Vue/Svelte components for the interactive bits. Clearly the right tool here.
  • •You want the nicest development experience → Svelte + SvelteKit. Less boilerplate, great docs, genuinely enjoyable. Svelte 5's "runes" reworked the reactivity model, so older tutorials may not match. Smaller ecosystem and hiring pool.
  • •Large team, long-lived enterprise app → Angular. Much better than its reputation now (signals, standalone components), and its strong conventions are an asset when 30 people touch the same codebase.
  • •You already have a backend (Rails, Django, Laravel, Phoenix) → consider not adding an SPA framework at all. Server-rendered HTML plus HTMX or Alpine.js handles a surprising amount, and you skip the whole API-layer tax.
  • •Middle ground between React and Svelte → Vue + Nuxt. Excellent docs, stable, less churn than React, healthy ecosystem — just fewer jobs and niche libraries in the US market.

Mostly settled regardless of choice: use Vite for tooling, TypeScript, and pnpm or Bun. Those are close to consensus now.

One caveat: this ecosystem moves fast and my knowledge has a cutoff, so check current release notes before committing — especially for Svelte, Solid, and anything Server Components related.

If you tell me what you're building, who's building it, and whether you need SEO or offline support, I can narrow this to one recommendation instead of six.

SEP 22afterPermalink

Picks read from the answer, in order
  1. blog
  2. Next.js
  3. Nuxt
  4. SvelteKit
  5. Angular

Short answer: if you have no strong constraints, React with Vite (plus React Router or TanStack Router) is the lowest-regret choice. Biggest ecosystem, easiest hiring, and every library and AI coding tool assumes it. But the better answer depends on a couple of questions:

Is it a content site — blog, docs, marketing?
Use Astro. It ships almost no JS by default, has great SEO/performance out of the box, and lets you drop in React/Vue/Svelte components for the interactive bits. Using a full app framework for a content site is the most common overkill mistake.

Do you need SSR, SEO, or server-side data fetching for an actual app?

  • •Next.js is the default and the safest for hiring. The App Router with Server Components is genuinely powerful but has a real learning curve and some sharp edges; budget time for it.
  • •Nuxt (Vue) is the pleasanter alternative if you don't need React specifically. Excellent docs, more conventional mental model.
  • •SvelteKit is the nicest DX of the three in my view — less boilerplate, less code to write — at the cost of a smaller ecosystem and hiring pool.

Is it a dashboard or tool behind a login, where SEO doesn't matter?
Skip the meta-framework. A plain SPA — Vite + React or Vue, TanStack Query for server state — is simpler and you'll avoid a lot of accidental complexity. This path is underrated right now.

Do you already have a Rails/Django/Laravel/Phoenix backend?
Consider not adding a framework at all. Server-rendered templates plus HTMX or Alpine.js handles more than people expect, and you keep one language and one deployment.

Large team, enterprise, want conventions enforced?
Angular is much better than its reputation — signals and standalone components modernized it considerably, and its opinionatedness is an asset at scale.

A few honest caveats. First, I'd avoid Solid or other newer options unless you specifically want to bet on them; they're technically impressive but you'll hit missing libraries. Second, my knowledge has a cutoff, so check current versions and release notes before committing — though framework-level churn has slowed a lot compared to the 2015–2020 era, so the landscape above is likely still roughly accurate.

Last thing: the choice matters less than it feels like it does. Architecture, data modeling, and knowing when not to reach for client-side state will affect your project far more than which of these you pick. Any of them can build a good app.

If you tell me what you're actually building, I can narrow it down.

Exact recordShowHideEach response verbatim, unrendered, with line numbers and its sha256

Before baseline

2026-09-21 permalink

Scored
answered
Received
2026-09-21 09:25:21 UTC
Run
20260921T090021Z-989a7c
Served by
Claude Platform on AWS · HTTP 200 · finish stop
Size
2,246 characters · 1,824 tokens out incl. hidden reasoning · 26.7 s
Extracted picks, in order · confidence medium
  1. Astro
  2. Svelte + SvelteKit
  3. Angular
  4. Middle ground between React and Svelte

Highlighted: the names the extractor took as picks, where they appear verbatim (4 of 4).

Short answer: **React with a meta-framework** is the safest default, but it's genuinely not the best pick for every project. Here's how I'd actually decide.
**Default choice: React + a meta-framework**
Largest ecosystem, most third-party components, easiest hiring, most Stack Overflow/LLM training data when you get stuck. For the meta-framework: Next.js if you want the most-traveled path, React Router (v7, formerly Remix) if you want something simpler and less coupled to a hosting vendor, or TanStack Start if you like that ecosystem. The main cost is complexity — React Server Components and the client/server boundary confuse experienced developers, not just beginners.
**Pick something else if:**
- **Content-heavy site** (blog, docs, marketing) → **Astro**. Ships almost no JS by default and lets you drop in React/Vue/Svelte components for the interactive bits. Clearly the right tool here.
- **You want the nicest development experience** → **Svelte + SvelteKit**. Less boilerplate, great docs, genuinely enjoyable. Svelte 5's "runes" reworked the reactivity model, so older tutorials may not match. Smaller ecosystem and hiring pool.
- **Large team, long-lived enterprise app** → **Angular**. Much better than its reputation now (signals, standalone components), and its strong conventions are an asset when 30 people touch the same codebase.
- **You already have a backend** (Rails, Django, Laravel, Phoenix) → consider not adding an SPA framework at all. Server-rendered HTML plus HTMX or Alpine.js handles a surprising amount, and you skip the whole API-layer tax.
- **Middle ground between React and Svelte** → **Vue + Nuxt**. Excellent docs, stable, less churn than React, healthy ecosystem — just fewer jobs and niche libraries in the US market.
**Mostly settled regardless of choice:** use Vite for tooling, TypeScript, and pnpm or Bun. Those are close to consensus now.
**One caveat:** this ecosystem moves fast and my knowledge has a cutoff, so check current release notes before committing — especially for Svelte, Solid, and anything Server Components related.
If you tell me what you're building, who's building it, and whether you need SEO or offline support, I can narrow this to one recommendation instead of six.

End of response · 18 lines · 2,246 characters · sha256 72b63d133379976d4233e642c24bea52acc6ccdda6da20e89b5e5d1c3839ed6e

After this run

2026-09-22 permalink

Scored
answered
Received
2026-09-22 09:25:12 UTC
Run
20260922T090022Z-7ff13b
Served by
Claude Platform on AWS · HTTP 200 · finish stop
Size
2,569 characters · 1,956 tokens out incl. hidden reasoning · 30.1 s
Extracted picks, in order · confidence medium
  1. blog
  2. Next.js
  3. Nuxt
  4. SvelteKit
  5. Angular

Highlighted: the names the extractor took as picks, where they appear verbatim (5 of 5).

**Short answer:** if you have no strong constraints, React with Vite (plus React Router or TanStack Router) is the lowest-regret choice. Biggest ecosystem, easiest hiring, and every library and AI coding tool assumes it. But the better answer depends on a couple of questions:
**Is it a content site — blog, docs, marketing?**
Use **Astro**. It ships almost no JS by default, has great SEO/performance out of the box, and lets you drop in React/Vue/Svelte components for the interactive bits. Using a full app framework for a content site is the most common overkill mistake.
**Do you need SSR, SEO, or server-side data fetching for an actual app?**
- **Next.js** is the default and the safest for hiring. The App Router with Server Components is genuinely powerful but has a real learning curve and some sharp edges; budget time for it.
- **Nuxt** (Vue) is the pleasanter alternative if you don't need React specifically. Excellent docs, more conventional mental model.
- **SvelteKit** is the nicest DX of the three in my view — less boilerplate, less code to write — at the cost of a smaller ecosystem and hiring pool.
**Is it a dashboard or tool behind a login, where SEO doesn't matter?**
Skip the meta-framework. A plain SPA — Vite + React or Vue, TanStack Query for server state — is simpler and you'll avoid a lot of accidental complexity. This path is underrated right now.
**Do you already have a Rails/Django/Laravel/Phoenix backend?**
Consider not adding a framework at all. Server-rendered templates plus **HTMX** or **Alpine.js** handles more than people expect, and you keep one language and one deployment.
**Large team, enterprise, want conventions enforced?**
**Angular** is much better than its reputation — signals and standalone components modernized it considerably, and its opinionatedness is an asset at scale.
A few honest caveats. First, I'd avoid **Solid** or other newer options unless you specifically want to bet on them; they're technically impressive but you'll hit missing libraries. Second, my knowledge has a cutoff, so check current versions and release notes before committing — though framework-level churn has slowed a lot compared to the 2015–2020 era, so the landscape above is likely still roughly accurate.
Last thing: the choice matters less than it feels like it does. Architecture, data modeling, and knowing when *not* to reach for client-side state will affect your project far more than which of these you pick. Any of them can build a good app.
If you tell me what you're actually building, I can narrow it down.

End of response · 24 lines · 2,569 characters · sha256 ea3d6943e4a4d1bb34c2e1d15e1bd00034b8374cd59ea4ca6d342565808689b6

Record detailsShowHideModel version, runs, baseline, triage, permalink, event ID
Model
anthropic/claude-opus-5
pinned anthropic/claude-opus-5-20260723; alias resolved to anthropic/claude-opus-5-20260723 at 2026-09-22 09:00:21 UTC (matches the pin)
Prompt
Question 31: P26 (recommendation)
Why this question
Tools for building websites change fast, so the answer shows how current a model’s knowledge is.
Change
first extracted pick differs from the baseline's (full lists below)
Baseline
6 prior days with data; 2 matched the baseline value (2026-09-18, 2026-09-21)
Responses
before 2026-09-21 09:25:21 UTC, run 20260921T090021Z-989a7c; after 2026-09-22 09:25:12 UTC, run 20260922T090022Z-7ff13b
Detected
2026-09-22 · run 20260922T090022Z-7ff13b, started 2026-09-22 09:00:22 UTC · scorer v1 · severity rules r2
Permalink
https://modeldrift.watch/e/2026-09-22/P26/anthropic/claude-opus-5/pick_change/picks.top/
Event ID
2026-09-22:P26:anthropic/claude-opus-5:pick_change:picks.top

Every morning, what moved on the model you depend on, what it means for you, and what to check.

$9 a month. Cancel any time. The next one tomorrow morning.

Get the daily brief

$99 once, for as long as modeldrift.watch runs. The first 100 buyers.

The record stays free. The brief is the reading of it.