Crimson AI NewsA CrimsonLingua Network service
EN ع
← Back to news
Research paper Hugging Face

Echoverse: Evolving Synthetic Environments Boost Computer-Use Agents from 36.5% to 67.1%

AI By Crimson AI Hugging Face Papers 1 August 2026 · 00:00 21 views
Share: X Telegram

Microsoft Research introduces Echoverse, a pipeline that compiles specifications into stateful synthetic applications with grounded grading, and a co-evolution loop that repairs environments and trains models. A 9B model trained on twelve environments jumps from 36.5% to 67.1% across fourteen evaluation splits, nearing a much larger frontier model.

Echoverse: Evolving Synthetic Environments Boost Computer-Use Agents from 36.5% to 67.1%

Key points

Computer-use agents learn from the consequences of their actions, so training them requires applications they can interact with, break, and reset. Real-world applications are often login-gated and stateful, making synthetic environments a practical stand-in. Recent pipelines generate such environments in bulk, shifting the bottleneck from quantity to the quality of what lies inside each one.

Researchers at Microsoft Research identify three properties that drive returns: behavioral depth, targeting the interactions where agents actually fail, and the ability to improve alongside the model. They present Echoverse, which compiles specifications into stateful applications whose tasks are graded against the application's own database. A co-evolution loop reads every graded rollout twice: once to repair the environment, its tasks, and its verifier, and once to provide training signal for the model.

Trained on twelve such environments, a 9B model improves from 36.5% to 67.1% across fourteen evaluation splits, coming within fourteen points of the much larger frontier model that taught it. The paper examines each property in turn: shallow environments can actually hurt live-site accuracy (dropping from 80.0 to 75.0), while deep ones raise it (from 80.0 to 85.0 and from 48.0 to 65.0). Drilling one interface control across many renderings transfers to held-out widget families and the open web. Repairing a single environment lifts a model trained on it from 16.2% to 38.5%.

The same worlds serve as reinforcement-learning environments. A reward combining the grounded verifier with a dense per-step judge raises held-out score from 58.8% to 68.0%. The authors release four environments as a benchmark, complete with applications, seed data, and grounded graders. Code and additional resources are available at the provided links.

MetricValue
Model size9B
Training environments12
Evaluation splits14
Improvement (base → Echoverse)36.5% → 67.1%
Gap to frontier model14 points
Shallow env live-site accuracy80.0 → 75.0
Deep env live-site accuracy (set 1)80.0 → 85.0
Deep env live-site accuracy (set 2)48.0 → 65.0
After single env repair16.2% → 38.5%
RL held-out score58.8% → 68.0%
Source
Hugging Face · Hugging Face Papers
Related news
Research paper
Hugging Face 31 Aug 2026

Hugging Face Unveils StepGuard: Step-Level Guardrails for Safer AI Agents

StepGuard, a new step-level guard model from Hugging Face, audits agent actions before execution, reducing attack success rates by...

1
Research paper
Hugging Face 31 Aug 2026

Hugging Face Researchers Unveil ABot-Recon for Stable Long-Horizon 3D Reconstruction

ABot-Recon, a new streaming 3D reconstruction model from Hugging Face, achieves stable long-horizon performance using only local t...

1
Research paper
Hugging Face 31 Aug 2026

ContextPilot: Teaching Agents Proactive Context Management via Fine-Grained RL

Hugging Face researchers introduce ContextPilot, a framework that enhances long-horizon agent reasoning by expanding context-editi...

1