Krux
February 16, 2026
Gradient's Echo-2 Slashes AI Training Costs by 80%
Published: February 16, 2026 at 4:47 AM
Updated: February 16, 2026 at 4:47 AM
100-word summary
Gradient launched Echo-2 in February 2026, a distributed reinforcement learning framework promising up to 80% lower post-training costs versus traditional cloud setups. Using a dual-swarm architecture that splits training and inference across distributed hardware, Echo-2 crushed benchmarks—reducing a 30B model's training time from 124 hours to 9.5 hours and costs from $4,490 to just $425. Early experiments show smaller models outperforming larger rivals. Gradient's upcoming RL-as-a-Service platform, Logits, opens reservations for researchers now, with enterprise access coming later in 2026. This could democratize cutting-edge AI development beyond Big Tech's data centers.
What happened
Gradient launched Echo-2 in February 2026, a distributed reinforcement learning framework promising up to 80% lower post-training costs versus traditional cloud setups. Using a dual-swarm architecture that splits training and inference across distributed hardware, Echo-2 crushed benchmarks—reducing a 30B model's training time from 124 hours to 9.5 hours and costs from $4,490 to just $425. Early experiments show smaller models outperforming larger rivals. Gradient's upcoming RL-as-a-Service platform, Logits, opens reservations for researchers now, with enterprise access coming later in 2026.
Why it matters
This could democratize cutting-edge AI development beyond Big Tech's data centers.