PrimeIntellect/INTELLECT-3

Text Generation • Updated Nov 27, 2025 • 2.5k • 210

Note INTELLECT-3 (Prime Intellect): A 100B+ parameter Mixture-of-Experts (MoE) model released in late 2025 and widely adopted in 2026. It is designed specifically for long-horizon tasks and can handle up to 10 million tokens by programmatically managing its own context.