DeepSeek's new open-source model reportedly crushes GLM and Kimi at 4-10x lower prices
anselm · x · 2026-09-11
Investor Deedy claims DeepSeek's new model may be the new king of open-source: it reportedly obliterates GLM 5.3 and Kimi K3 on benchmarks while being 4-10x cheaper, priced at $0.3/M input and $1.2/M output, and ranks #6 among humans on Codeforces. Numbers are unverified pending official confirmation.
Scoble adds a geopolitical angle: Sand Hill money is watching China tonight, and he argues the nearer AI risk isn't long-term but infrastructure security — a better-funded AI lead could let bad actors do massive harm, so the US and China should talk to ensure no one gets an overwhelming advantage.
Related event: Leaked DeepSeek-V4.1-Flash Touted as New Open-Source King(3 posts)→
More from AGI Musings
- Ben Todd: Many researchers put AI existential risk at 30%-70%, well above 10% — ben_j_todd · 2026-09-11
- Ex-Anthropic researcher Jacob Coxon on NBC: recursive self-improvement is the warning sign to watch — rohanpaul_ai · 2026-09-11
- Musk predicts AI will surpass all human intelligence within 5 years — JRIngallinera · 2026-09-11
- tszzl: plain-text CoT observability will look like an alchemical era — tszzl · 2026-09-11
- Doomerism is a category error: aligning arbitrary AI vs specific AI — inductionheads · 2026-09-11
- Derek Thompson: Not Everything in the AI Safety Debate Is a Psyop — nptacek · 2026-09-11