A total disaster

Overview

The speaker argues that OpenAI’s “DevDay” rollout was widely hyped as exciting, but quickly became controversial due to behind-the-scenes changes and perceived over-marketing of existing ideas.

Key Points and Criticisms

1) Usage limits were quietly reduced right before DevDay

• The day before the event, OpenAI reportedly tweeted that usage limits would be cut in half, while prices stayed the same.
• This allegedly led to backlash and confusion among users.

2) New products are portrayed as copycat or derivative

The speaker claims DevDay introduced tools that resemble competitors, including:

• “Copies” of Muse
• Gev/JeV/Grok-style agents
• A Notion-like product (“Spaces”)

Overall, the allegation is that OpenAI repackages existing products rather than delivering genuinely new breakthroughs.

3) Plans and pricing changes are criticized as misleading

• A previously “best” plan (around $200/month) allegedly saw usage drop from 20x to 10x.
• OpenAI allegedly responded by claiming the models were “twice as effective.”
• The speaker dismisses the supporting evidence (“bro benchmarks”) as dubious.
• They also argue OpenAI can change scope/terms after payment, which feels unfair.

4) A high-tier plan is framed as a “speed upgrade” with heavy cost

• The speaker discusses a $500/month plan with “Super fast mode.”
• The claim is 8x faster, but 6x more expensive.
• They demonstrate code-edit speed on a “V2 migration” task, noting they consumed only a small portion of their weekly limit—suggesting the speed is highly tempting (or “addictive”).

5) “Unlimited” marketing is reframed as limited-duration and conditional

• The speaker addresses claims that a product (“Dots”) offers unlimited usage.
• They argue it’s actually unlimited for only one month.
• They also claim usage may still incur costs if the system delegates work to other agents.

6) Specific product delays and logistics are mocked

• The speaker criticizes an “announcement of an announcement,” claiming an image-capable decision model was unavailable for about a week after the claim.
• They also dismiss a supposed “Wi‑Fi excuse” during a demo attempt, implying the demo simply didn’t work as shown.

Token Sharing via ChatGPT Login (Praise)

The speaker highlights what they see as a meaningful improvement:

• Logging in via ChatGPT so third-party apps can share/charge users’ tokens without exposing API keys.
• The approach may also support spending limits for safety.
• They predict this model could become common and widespread.

Questioning “Openness” Messaging

• The speaker references OpenAI promoting openness (for example, availability of certain models in Codex, with costs counting toward limits).
• They argue “openness” may be strategic rather than truly empowering users/control.
• The suggestion is that OpenAI may leverage openness claims while keeping key components closed.

Overall Takeaway

The speaker concludes that DevDay felt less like a visionary leap and more like:

• pricing/limit tightening, plus
• repackaged competitor features,

with notable exceptions being:

• token-sharing integration through ChatGPT, and
• the appeal of much faster coding/agent workflows—though at significant cost.

Presenters / Contributors Mentioned

• OpenAI (company)
• Sam (referenced as having said new stuff is coming)
• Sarah Fryer (CFO of OpenAI)
• Thibaut (referenced via tweets)
• T-Bone (referenced via tweet)

Other Entities / Models / Companies Mentioned

• SpaceX
• Cloudflare
• Anthropic (as comparison)
• GLM / “Kimi” (referenced as upcoming availability)
• Codex
• Cursor
• JeV/Grokbot/Muse/Notion-like tools (as comparisons)