Latest

Developers using LLM agents for coding can reduce errors and costs by 14-19% without retraining, transferring across models and benchmarks.

ACE improves efficiency for MoE models without retraining, benefiting developers working with large language models in resource-constrained environments.

Developers can use Terminal-Universe to create more realistic training environments for terminal-based code agents, improving their performance on real-world coding tasks.

This quantization method enables smaller, faster hybrid LLMs without significant performance loss, making large models more accessible for resource-constrained environments.

Developers can implement SafeEvolve to improve agent safety without sacrificing utility, achieving better safety-utility tradeoffs than existing methods.

MT-SDPO improves multi-domain LLM performance by 14.79 points on Qwen3-8B, offering better domain balance than single-teacher approaches.

Qwen3.8-Max-0902 offers top coding performance at competitive pricing compared to alternatives like Claude Opus 5 and Kimi K3.

官方账号阿里通义 Qwen@Alibaba_QwenSource

Global quantization granularity beats selective layer repair by up to 52 accuracy points, offering a better default for LLM deployment.

Released under Apache-2.0 license with lowest median latency at 14.3 ms, offering robust protection against Russian and English prompt injections.

The study provides concrete guidance for selecting LLMs in ontology engineering, showing model architecture matters more than size for specific tasks.