OpenAI mulls slashing prices as it competes with Anthropic for users

Original Article ↗ Hacker News Discussion ↗

Profitability & Business Models

  • Many doubt how frontier labs become profitable when training and infra dominate costs and models are sold below cost (“selling a dollar for 50 cents”).
  • Some argue inference is profitable and R&D is the main drag; others say you can’t ever “stop training” without falling behind, unlike Amazon pausing expansion.
  • Comparisons are made to Uber-style market-share subsidization, not early Amazon’s reinvested profits.
  • Some think the real goal is a big IPO and equity payout rather than sustainable profits.

Price Cuts and Competitive Dynamics

  • Users see a “race to the bottom”: cutting prices to gain share but deepening losses, risking a bubble and potential bankruptcy.
  • Some think OpenAI is trying to drag Anthropic into a cash-burn war it can’t win.
  • Others note Anthropic has raised prices on top models (e.g., Fable, Opus variants) even as cheaper competitors like DeepSeek cut prices.

Model Quality & Usage Experience

  • Split views: some find Claude/Fable clearly superior for coding and long “agentic” tasks; others say Codex/GPT 5.5 are now better or at least “good enough.”
  • Token limits are a recurring pain point, especially with Anthropic’s subscription + per-use model; some users never hit limits, others hit them frequently.
  • Cheaper models like DeepSeek and Kimi are praised for ultra-low cost but criticized as slower, less reliable, and not yet safe to run unsupervised.

Enterprise vs Consumer Focus

  • Many think Anthropic is more focused on pros/enterprise, while OpenAI chases both mass consumer and enterprise, seen as strategic diffusion.
  • Debate over whether enterprise (pay-per-token, heavy usage) or consumer (subscriptions, huge user base) is ultimately more lucrative.

Local/Open Models & Commoditization

  • Several expect LLMs to become a commodity, with many users eventually running “good enough” open models locally or on cheap rented GPUs.
  • Chinese providers (DeepSeek, Qwen, GLM) are cited as existential price competitors, especially outside the US.

Ethics, Trust, and Company Perception

  • A number of commenters avoid OpenAI over perceived warmongering, ad ambitions, or leadership hypocrisy and are happy to see it lose ground.
  • Others argue they’ll choose tools based on capability and cost, but ethics can easily tip decisions when quality is “good enough.”

Bubble / Long-Term Outlook

  • Some foresee an AI bubble akin to dotcom: LLMs survive, but current leaders may not (AOL/Yahoo analogy).
  • Others think AI will settle as a must-have developer and cybersecurity tool, with expectations scaled back but industry enduring.