What the internet is saying about OpenAI o3

What the internet is saying about "OpenAI o3"

Caveat: None of the supplied snippets names o3 directly. The evidence is about OpenAI broadly, plus developer work on "reasoning models" and reasoning_effort, which is the closest o3-adjacent signal. Treat the o3-specific conclusions below as inference.

🟢 Supporters

• •Reasoning models are being built into developer tooling. PraisonAI PRs send reasoning_effort as reasoning.effort on the Responses API [2], drop sampling params for reasoning models [3], and use max_completion_tokens for them [4]. Silo-Server added per-model reasoning levels and browser ChatGPT sign-in [6].
• •OpenAI's distribution and tooling are expanding. "OpenAI models are coming to Google Cloud" drew 326 points [10]. The "OpenAI CLI" post drew 244 points [11]. "Scaling Git at OpenAI" drew 271 points [7].
• •OpenAI is making acquisitions. The Pierre Computer Company deal drew 263 points [9].
• •Codex and ChatGPT are used in real agent workflows. Examples include Codex seat coordination [20], ChatGPT × Arena handoffs [17], and a multi-model "firm of agents" that lists Codex and ChatGPT as workers [36].
• •Prediction markets show sustained interest. There are active Polymarket markets on an OpenAI IPO, its closing market cap, and another Millennium Prize solution [12–15]. Volumes range from about $142K to $317K.

🔴 Skeptics

• •Funding: The WSJ headline says OpenAI raised less than half of its $30B target (188 points, 203 comments) [1].
• •Privacy and trust: A post claims OpenAI is silently re-enabling model training on Codex chats. It drew 526 points, and one commenter says their opt-out was "re-enabled" despite two granted requests [8].
• •Security and incidents: An OpenAI engineer was arrested over bugging an executive conference room (657 points, 545 comments) [18]. Fortune reports four more "rogue incidents" [21].
• •Product limits: "I Am the Reason OpenAI Is Killing Unlimited Chats" suggests usage caps are coming [16].
• •Ethical and competitive pushback: "Just say no to OpenAI and Anthropic" [23], "Why Math Dumped OpenAI" [26], and "The math that OpenAI keeps for itself" [27].
• •Market sentiment: Several Polymarket markets are trending down: an IPO market-cap market is down 25.7% this month [12], the Millennium Prize market is down 10.5% [14], and the IPO-by market is down 2.2% this week [15]. The snippets don't say what each percentage measures.

⚪ Neutral Observers

• •Much of the reasoning-model GitHub activity is compatibility plumbing: parameter handling, token naming, and effort-level mapping. It is not praise or criticism of model quality [2][3][4].
• •Silo-Server mapped effort levels for Grok and DeepSeek because those APIs don't accept every level. This shows reasoning-effort controls are spreading across vendors [5].
• •Maxim's Bifrost updated tests because some provider models were retired or deprecated [22]. This is a reminder of how fast model lineups change.
• •Many GitHub items mention ChatGPT or Codex only incidentally, as part of agent or governance workflows [17][19][20][35][36].

Cross-Platform Signals

• •Hacker News is the most active platform and the most critical. Its top items are about trust, governance, funding, and incidents [1][8][18][21].
• •GitHub is mostly practical. Developers are integrating and adapting to reasoning-model APIs, with little opinion either way [2–6].
• •Polymarket shows speculative interest in the IPO and in math breakthroughs, with mixed-to-falling momentum [12–15].
• •Divergence: Developers keep building on OpenAI's reasoning stack while HN discussion about the company is largely negative. No source discusses o3's benchmark or quality directly.

Community Verdict

The evidence doesn't support a clear verdict on o3 itself. It shows OpenAI as widely used and still expanding, through reasoning-API integrations, Google Cloud availability, and a new CLI. The surrounding discussion is skeptical, centered on a funding shortfall, training-data and privacy concerns, security incidents, and usage limits. Developers treat reasoning models as standard infrastructure, while the wider conversation questions OpenAI's trustworthiness and finances more than its model capabilities.

---

Based on Hacker News + GitHub + Polymarket data from the last 30 days, researched with last30days by Matt Van Horn.