OpenAI

The San Francisco AI lab (founded 2015, restructured through several capped-profit and nonprofit-parent arrangements since) that trained and shipped GPT-3, GPT-3.5, GPT-4, and the o1 reasoning-model series, and that runs the developer API much of the rest of the industry builds on top of.

Matters to this vault not as a byline but as an actor across more than one independent cluster of notes:

• The o1 chain-of-thought decision. OpenAI's own primary launch page states that it decided not to show users o1's raw chain of thought "after weighing multiple factors including user experience, competitive advantage, and the option to pursue the chain of thought monitoring" — the company naming a strategic, IP-protective reason for a design choice in its own words, not an observer's inference (claim-openai-hid-o1-raw-chain-of-thought-partly-for-competitive-advantage). What it ships instead is a model-generated summary of the reasoning, also its own stated choice (claim-openai-o1-shows-model-generated-summary-of-chain-of-thought-not-raw-trace).
• Embedding API pricing. OpenAI cut its embedding API list price 5x across one model generation, from text-embedding-ada-002 to text-embedding-3-small ($0.10 → $0.02 per 1M tokens) — a direct pricing decision, not a third party's report about one (claim-openai-embedding-price-fell-5x-ada-002-to-3-small).

References

• claim-openai-hid-o1-raw-chain-of-thought-partly-for-competitive-advantage
• claim-openai-o1-shows-model-generated-summary-of-chain-of-thought-not-raw-trace
• claim-openai-embedding-price-fell-5x-ada-002-to-3-small