Claude Max
$100 / $200 per mo (Pro $20)
Models: Claude only (incl. Claude Code)
What "flat" means: Fixed price with metered usage: 5-hour session windows plus a weekly cap; Max raises them 5x or 20x over Pro and adds $100 or $200 of API credits a month.
The catch: Session amounts aren't published, and the plan login only works in Claude Code and Anthropic's apps (other agents get only Max's API credits). Claude models only; heavy agentic use hits the windows.
Pick Claude Max when
+Claude Code is your daily driver; it remains the quality benchmark among coding agents
+Your usage fits inside the session windows and weekly caps
+You want chat, projects, and the agent bundled with zero API setup
Nous Portal
$20 / $100 / $200 per mo
Models: Hundreds of models (tier cards say 200+, the catalog 300+): Claude, GPT, Gemini, DeepSeek, Qwen, Kimi, MiniMax, Grok…
What "flat" means: Fixed price buying 1.1x its value in credits ($22 / $110 / $220), with unused credit rolling over up to $10 / $50 / $100, plus a tool gateway (web search, scraping, image generation, browser use, speech) priced per use from the same credits.
The catch: A ~1.1x credit multiple, and tools draw on the same credits: value comes from Hermes-native auth and breadth, not raw inference discount. An eligible ChatGPT plan can be connected for inference instead.
Pick Nous Portal when
+You run Hermes Agent and want its native subscription with one auth flow
+You value managed tools (web search, image, speech, browser) billed from the same credits as inference
+Broad frontier + open model access matters more than lowest cost-per-token