AI

Former OpenAI and Anthropic researcher Jacob Coxon says Anthropic's IPO graph shows superexponential gains in AI training efficiency, claims Mythos could already be retrained with a fraction of its original compute

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 14/11/2026, 15:47:26 UTC.

Briefings

A former OpenAI and Anthropic researcher claims Anthropic's IPO data indicates AI training efficiency gains that are superexponential, suggesting the company could theoretically retrain models with a fraction of its original compute. This claim is based on a single Reddit post from November 2026, which lacks peer review or independent verification.

Key points

• Jacob Coxon, a former OpenAI and Anthropic researcher, publicly stated that Anthropic's IPO graph demonstrates superexponential gains in AI training efficiency.
• Coxon claims this efficiency allows for model retraining with a fraction of the original compute, though no independent data supports this assertion.

Context

This finding originates from a single Reddit post by a former employee in November 2026. No additional background, expert analysis, or corroborating evidence is supplied in the source material.

Limitations

Evidence is limited to one Reddit post with no peer review or independent verification. The claim relies entirely on unverified data from a former employee without corroborating evidence.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 14/11/2026, 15:48:36 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeihvq75dst5rgfuimkvhymnuxqusfceq35443dreybbbz75muwekjy