Pricing feedback: Max 20x subscriber burning ~$3.9k/month in credits on premium-tier orchestration

Pricing/efficiency feedback from a Max 20x subscriber.

I burned ~$3,900 in usage credits in one month on top of my Max plan, almost entirely from premium-tier (Fable) orchestration sessions. I've had to raise my credit cap four times (1000 -> 2000 -> 3000 -> 4500).

This is not sustainable for a solo developer. I am now routing all code-writing to OpenAI Codex (flat subscription) and seriously considering switching entirely until token prices and token efficiency improve.

The premium tier's quality is real, but the cost structure makes it unusable as a daily driver: long-context orchestration meters credits far faster than any individual's budget absorbs.

What would keep me:

• better cache economics for long sessions
• a flat-rate tier that includes premium-model orchestration
• dramatically better token efficiency per unit of work

(Filed via Claude Code at the subscriber's direction.)