MiMo near-SOTA on DeepSWE with just ~$2.6M RL run: will data cost more than training?
my_cat_can_code · x · 2026-09-21
mycatcancode raises a structural question about training economics: might data costs soon exceed compute costs for model training?
The evidence: MiMo achieved near-SOTA results on DeepSWE with an RL run costing only $2.6M. As RL training gets dramatically cheaper, acquiring and using high-quality data could become the dominant expense.
The post frames this as a potential shift in the training cost structure — compute getting cheaper while data gets more expensive.
Related event: $2.6M RL run nears SOTA, sparking training-cost debate(4 posts)→
More from Infra
- Leaked memo: Meta infrastructure head Santosh Janardhan to leave in early 2027 — dinabass · 2026-11-12
- Tool-call failure rates jump 14x across providers, charts pinpoint inference stack updates — remilouf · 2026-11-12
- Omarchy 4.6 lets you share local AI models across your tailnet, says DHH — alexcovo_eth · 2026-11-12
- Eric Schmidt's foundation funds French NGO fighting against data centers — StephanSturges · 2026-11-12
- Epoch: Open-Weight GLM 5.3 Hacks Six Firefox Versions for ~$1,000 Each; Claude Fable 5.2 Sets ECI Record — Epoch AI · 2026-11-12
- Polymarket prices 28% odds of a state data center moratorium; Torus raises $500M — Polymarket · 2026-11-12