Endless Frontier releases Fx-Bio-1031 bioinformatics model
Endless Frontier has released the full weights of Fx-Bio-1031, a bioinformatics model designed for professional tasks such as analyzing multi-omics data, writing analysis scripts, querying databases and validating results. The model is based on Qwen3.8-Flash-Next, supports tool use and local sandbox deployment, and has a native 512K context window. QbitAI reports that, in a team evaluation framework, Fx-Bio-1031 scored 70.59% Pass@5 on the Difficult subset of Anthropic's BioMysteryBench, compared with 47.06% for GPT-6-Astra and 52.94% for Claude Opus-5. It also recorded 89.50% Pass@3 on BiomniBench. The report attributes the comparison chart to a Hugging Face official evaluation and separately describes an internal independent validation system.
Full-weight release, local deployment and tool-oriented bioinformatics workflows give research teams and developers control over model deployment and data handling. The reported benchmark results show the model’s claimed performance on difficult scientific tasks, although the published figures are presented as evaluation results rather than proof of broad real-world effectiveness.
高难盲测超越GPT-6-Astra与Opus-5!无界前沿开源Fx-Bio-1031,把RSI带进生命科学 – 量子位
< img id="wx_img" src="https://www.qbitai.com/wp-content/uploads/imgs/qbitai-logo-1.png" width="400" height="400">
高难盲测超越GPT-6-Astra与Opus-5!无界前沿开源Fx-Bio-1031,把RSI带进生命科学

数据不出域、本地可部署
#Fx-Bio-1031#Endless Frontier#Qwen3.8-Flash-Next#RSI-OS#Anthropic BioMysteryBench#BiomniBench#GPT-6-Astra#Claude Opus-5#BioMaster#Trisol#Deep Potential