Nexusflow’s latest AI model outperforms GPT-4 in software tool usage
Generative artificial intelligence cybersecurity startup Nexusflow.ai Inc. today announced the release of NexusRaven-V2, a 13 billion-parameter open-source generative artificial intelligence model that delivers function calling capability and is claimed to outperform GPT-4 in software tool usage.
Function calling allows the large language model to understand human instructions and translate the instructions into precise function/application programming interface calls to use in various software tools. The capability lies at the core of the OpenAI Assistants API and serves as the key to enabling copilots and agents to use software tools.
The new model, instruction-turned from the CodeLlama-13B model, achieves up to a 7% higher tool use success rate than the latest OpenAI GPT-4 model on human-curated general software tool use benchmarks.
The data curation pipeline for the new model combines open datasets and Meta’s Llama ecosystem to curate massive amounts of high-quality data, which was used to train NexusRaven-V2.
The data curation and training of NexusRaven-V2 did not involve proprietary LLMs, such as OpenAI’s GPT-4, allowing enterprise customers to own and use the model to build copilots and agents. Nexusflow notes that this is especially important for enterprise applications that require up-to-date information, quality, safety and in-depth domain customization.
According to Nextflow, NexusRaven-V2 offers superior quality and has the potential to revolutionize workflow automation on complex software with a significantly smaller model size and higher accuracy.
Nexusflow’s customers can own their generative AI model without data risks associated with proprietary model vendors. The company says that with the release of NexusRaven-V2, customers can enjoy superior generative AI quality, reliability, cost reduction and safety with complete and up-to-date customization, compared to solutions built on proprietary model vendor APIs with unpredictable quality degradation and lagging knowledge.
“NexusRaven-V2 outperforms OpenAI’s GPT-4 model head-to-head for using software tools,” co-founder and Chief Executive Jiantao Jiao (pictured) said ahead of the release. “This validates Nexusflow’s technical capability to deliver enterprise solutions using open-source models.”
Nexusflow is a venture capital-backed startup, having last raised $10.6 million in September. The round was led by Point72 Ventures LP, with Fusion Fund LLC and a number of individual investors also participating.
A message from John Furrier, co-founder of SiliconANGLE:
Your vote of support is important to us and it helps us keep the content FREE.
One click below supports our mission to provide free, deep, and relevant content.
Join the community that includes more than 15,000 #CubeAlumni experts, including Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger, and many more luminaries and experts.