Kimi V4 is headed for native multimodality, but Moonshot says compute is the limit
teortaxesTex · x · 2026-07-23
Moonshot’s Kimi V4 is said to be heading toward native multimodality, and the image-backed quote says V4 is already as large as the team can train right now.
Key points from the post and image:
- Kimi V4 and its follow-up versions are expected to support native multimodality.
- The model is described as being trained at the largest size the team can currently afford.
- The implied bottleneck is compute, not lack of willingness to scale.
- The post also suggests that a successful funding round could let Moonshot train larger models and push further on scaling.
The broader message: Kimi is signaling both a multimodal roadmap and a very explicit compute constraint.
More from Infra
- SF Compute founder: buying compute is 'an absolutely awful experience' right now — IgorCarron · 2026-09-11
- SmolVM open-sources persistent computer infrastructure for agents that outlive chat sessions — aniketmaurya · 2026-09-11
- PyTorch Day Korea 2026 launches first offline conf, CFP closes Sept 13 — PyTorch · 2026-09-11
- Local LLM server dilemma: 4x CMP-170HX (price up 53% in 20 days) vs Mac Studio M5 Ultra — rumboll · 2026-09-11
- llama.cpp lands Flash Attention tuning for RDNA4, big prefill gains on AMD — pmttyji · 2026-09-11
- Your p99 latency benchmark may be lying: a deep dive into coordinated omission — Franc0Fernand0 · 2026-09-11