Ex-Anthropic researcher says retraining Mythos needs only a fraction of original compute
11/14/2026 — 11/17, 12:17·2 sources (all)·2 reports (all)
Key takeaways
- Jacob Coxon, a former researcher at OpenAI and Anthropic, said an IPO roadshow chart showing super-exponential gains in algorithmic efficien…
Background & analysis
On November 14, 2026, IT Home reported that Jacob Coxson, a researcher who previously worked at both OpenAI and Anthropic, said that Anthropic's Mythos model, released this spring, can now be retrained using only a small fraction of the compute required for its original training run. Coxson, who has drawn wide attention for publicly warning about AI risks, spent only about four months at Anthropic, and he cited a slide Anthropic presented during its IPO roadshow as a key reason for leaving the company.
According to Coxson, that slide illustrated how the compute demands of training AI models have shifted. It used a compressed axis, so even 10x or 100x growth in compute requirements appears visually modest, yet the curve still rises steeply. He argued this indicates that algorithmic efficiency is not merely improving quickly, but that the pace of improvement is accelerating, and he therefore concluded that Mythos's performance breakthrough was not an accident, but a node in a trend of super-exponential growth. Coxson added that algorithmic improvements are approaching automation, and he called for a pause in frontier AI model development next year, which would give competitors a chance to catch up. He also argued that such a step would make the arrival of human-level intelligence attainable.
AI-generated from 1 reports · updated 3 days ago
Latest turnAccording to Tom's Hardware, Anthropic told investors that Claude Mythos Preview got its AI running on AMD GPUs in about three days with virtually no human intervention, compared to the usual 6 to 12 months.
Related AI tools
- Heat index
- 98
- All sources
- 2
- All reports
- 2
- First seen
- 3 days ago
Reports on this story headlines open the original
According to Tom's Hardware, Anthropic told investors that Claude Mythos Preview got its AI running on AMD GPUs in about three days with virtually no human intervention, compared to the usual 6 to 12 months.
TechmemeAI score 59
Jacob Coxon, a researcher who has worked at OpenAI and Anthropic, said a compute chart shown in Anthropic's IPO roadshow helped prompt his departure, arguing it shows algorithmic efficiency improving at a super-exponential pace. According to IT之家, he said retraining Mythos now takes only a fraction of the original compute and called for a pause on frontier model development next year.
IT之家AI score 53
Other stories people are talking about
- Heat index 80Open-source Agent Keychain lets agents use credentials without seeing them1 sources
- Heat index 97NewDeveloper Launches Unauthenticated MCP Service TLDMint for Domain Availability1 sources
- Heat index 1336RisingNvidia board not discussing Jensen Huang's return, probe into his conduct advances22 sources
- Heat index 193SurgeClaude Fable 5.2 gets a 20% price cut, but per-task cost rises at max effort4 sources
- Heat index 108Claude Code Code Mode can be enabled via an environment variable2 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

