ChatGPT Plus / Pro
$20 / $100–500 per mo
Models: OpenAI only (incl. Codex)
What "flat" means: Fixed price. Plus has a five-hour window (an estimated 15 to 160 GPT-6.1 Sol messages) and a weekly limit; Pro has no five-hour limit but still caps usage weekly. Codex local and cloud tasks share one allowance.
The catch: OpenAI models only, and weekly limits aren't published; parallel Codex tasks drain the shared allowance fast. Other agents can use the plan only through the Sign in with ChatGPT preview for open-source apps.
Pick ChatGPT Plus / Pro when
+You want the ChatGPT product itself alongside the coding agent
+OpenAI's cloud Codex tasks and managed sandboxes matter to your workflow
+OpenAI's GPT-6 models are your preference and the plan limits fit your volume
Featherless
$25 (Chat) / from $50 (Developer) per mo
Models: Open-weight only (DeepSeek, GLM, Kimi…)
What "flat" means: Chat plan: flat price with uncapped tokens, but for chat typed by you. Developer plan: monthly credits billed per token.
The catch: The flat-price Chat plan excludes coding agents and API traffic (32K context, 4 concurrent units); coding agents need the per-token Developer plan (up to 256K context).
Pick Featherless when
+Open-weight models genuinely cover your quality needs
+You chat with open-weight models yourself and want uncapped tokens at a low flat price
+Model pinning on the open-weight universe matters to you