ALL MEMOS Download .docx

Sonnet-only mode — pros, cons, best utilization

TL;DR

Sonnet is the workhorse. For 90% of TITAN work, you lose almost nothing by running Sonnet-only for the rest of the week. Save the deep philosophical / synthesis / new-strategy work for after Opus resets. Push everything else through Sonnet.

What model is Sonnet anyway

Claude Sonnet 4.5 is Anthropic's mid-tier flagship. As of late 2025/early 2026:

For most TITAN ops, Sonnet is functionally indistinguishable from Opus. You'd have to side-by-side them on a hard problem to notice.

PROS of running Sonnet-only this week

1. Speed — Sonnet replies 30-50% faster than Opus. Your iteration velocity increases.

2. Cost — if API is involved (PostCompact extractor, embedder, agent dispatches), Sonnet costs less per token.

3. You keep working through the cap — instead of being blocked, you ship.

4. Code quality is unchanged — Sonnet is actually preferred for code by many senior engineers.

5. Forces leverage discipline — knowing Opus is precious, you save it for the work that genuinely needs it. This is a habit upgrade.

CONS of running Sonnet-only this week

1. Deep synthesis memos — the PhD-depth SCOUT memos (A074 spiritualaf.me, A079 memory architecture) are Opus-territory. Sonnet versions will be ~85% as good but missing some of the subtle pattern-finding. Defer these until reset.

2. Subtle voice matching — when matching your specific lion-tiger / chill-flow voice with high precision (e.g. drafting your speaking-page bio), Opus catches nuance Sonnet sometimes misses. Save voice-critical work.

3. Multi-step planning across 10+ files — Sonnet can do it but tends to drift more on very long horizons. Keep big architectural decisions for after reset.

4. Some cron skills are configured to use Opus — they auto-fall-back to Sonnet, but check the model: line in each SKILL.md.

Best utilization plan for the rest of this week

USE SONNET FOR (push it hard):

DEFER UNTIL RESET (Opus territory):

CONFIGURE NOW

When does the cap reset?

Anthropic's weekly fair-use band typically resets at the start of your subscription's billing week. Look at your Settings → Billing to find the exact day. Most users see it reset Sunday-evening UTC.

The deeper move

The cap is a feature, not a bug. It forces leverage thinking:

The men who outscale others in the AI era are the ones who route the right model to the right task. You're learning that this week. Use it.

The single sentence

Sonnet-only does 90% of TITAN's work without losing meaningful quality. Push everything except the deep-synthesis memos and the voice-critical singleton work. Reset comes Sunday-ish; until then, the lion still hunts — just on the slightly leaner cat.

— TITAN, with mass love · 2026-04-27