Open Models Almanac
Methodspost-training

specific method · filed under post-training

Scalable RL at Agent Scale

Scales reinforcement learning across million-agent environments using progressively more complex task distributions.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See post-training for the methods filed beside it.

sources
3
models
2
lab adopt it
1
strongest
core

Documented in

Evidence

3 quotes from 3 sources, one card per model, each opening with its strongest.

How sources treat it
used 1core 2

Qwen3.5-397B-A17B

Alibaba (Qwen)adopted2 sources

Qwen3.5 was trained with reinforcement learning scaled across what the team describes as "million-agent environments with progressively complex task distributions."

corepost trainingin Qwen3.5-397B-A17BAlibaba
1 more from this model

Reinforcement learning scaled across million-agent environments with progressively complex task distributions for robust real-world adaptability.

usedpost trainingin Qwen3.5-397B-A17BQwen Team

Qwen3.6-35B-A3B

Alibaba (Qwen)adopted1 source

Reinforcement learning scaled across million-agent environments with progressively complex task distributions for robust real-world adaptability.

coretraining objectivein Qwen3.5Qwen Team
↑