Model Releases
Impressive by @river_ai_inc to beat @deepseek_ai rates - cache in particular is tough! Talented team there, full RL + inference stack should…
Impressive by @river_ai_inc to beat @deepseek_ai rates - cache in particular is tough! Talented team there, full RL + inference stack should be real competition River Inference is live! We're serving
Impressive by @river_ai_inc to beat @deepseek_ai rates - cache in particular is tough! Talented team there, full RL + inference stack should be real competition River Inference is live! We're serving DeepSeek V4.1 Flash and GLM 5.3 Flash for 20% below official provider rates.
Related
- Congrats to @deepseek_ai team! Doing the numbers I would estimate: Pro < 14m for the final training run Flash < 4m Ratio of active params …
- Necessity is the mother of kv-cache optimisation
- Metronome: Bound the Cache, Keep the Beat for Real-Time Interaction Model Serving
Source: Emad Mostaque (X) | 2026-10-11