VQV Signal
According to a post on X by @QwenAIFans, the Beijing Academy of AI (BAAI) has open-sourced AREX-Turbo, a deep-research agent built on Qwen3.5-4B with a 260K context window. The post reports that the model scores 70.7 on BrowseComp, above Tongyi DeepResearch 30B's score of 43.4. It performs recursive evidence search and is small enough to be served locally with vLLM.
According to a post on X by @QwenAIFans, the Beijing Academy of AI (BAAI) has open-sourced AREX-Turbo, a deep-research agent built on Qwen3.5-4B with a 260K context window. The post reports that the model scores 70.7 on BrowseComp, above Tongyi DeepResearch 30B's score of 43.4. It performs recursive evidence search and is small enough to be served locally with vLLM.
AI-assisted summary based on the listed source.
The Beijing Academy of AI's model, built on Qwen3.5-4B with a 260K context, reportedly scores 70.7 on BrowseComp, above Tongyi DeepResearch 30B's 43.4. It does recursive evidence search and is small enough to serve locally with vLLM. Original post on X Section: Organization News
Developers and teams building research or web-search agents may find a 4B model that can run locally a smaller-footprint option to compare against much larger open models. The BrowseComp score is the key figure to check, since the source describes it as reported and it has not been independently confirmed here. Watch for independent benchmark results and whether the released model matches the claims.
VQV organizes public signals from inspectable sources. It does not independently verify the underlying report.
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
VQV surfaced this signal because it is recent, relevant to Open Source LLMs, connected to Scobleizer News Pulse (X).
No login, cookies, social SDKs, or automatic posting.
