NexusRaven Models by Nexusflow
About
The NexusRaven family of large language models (LLMs), developed by Nexusflow, is designed for efficient function calling within specialized software tools. The initial model, NexusRaven-13B, is renowned for its impressive 95% success rate when interacting with cybersecurity tools such as CVE/CPE Search and VirusTotal, significantly surpassing GPT-4's 64% success rate 13. NexusRaven models excel in generalizing to unseen tools, performing on par with GPT-3.5 in zero-shot scenarios 1. Their training is distinct, avoiding proprietary LLM-generated datasets, which adds to their commercial appeal 1. Continuing advancements are evident in subsequent versions like NexusRaven-V2 2. The models are available in various quantized formats, such as GGUF and GPTQ, making them adaptable for different hardware and inference tasks, thus broadening their applicability across platforms 3.
Current Variants
Use-when guidance is derived from seed capabilities, context, release, and replacement fields.
Use when the workload needs 2k context and 13B parameters.
Use when the workload needs 100k context, 13B parameters, and structured outputs.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| NexusRaven 13B | Use when the workload needs 2k context and 13B parameters. | 2023-10 | 2k context13B parameters | Current |
| NexusRaven-V2 13B | Use when the workload needs 100k context, 13B parameters, and structured outputs. | 2023-10 | 100k context13B parametersstructured outputs | Current |
Release Timeline
1 release groupSpecifications(2 models)
| Model | Released | Context | Parameters | Structured Outputs |
|---|---|---|---|---|
| NexusRaven 13B | 2023-10 | 2k | 13B | No |
| NexusRaven-V2 13B | 2023-10 | 100k | 13B | Yes |
Available From(1 provider)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| NexusRaven-V2 13B | Together AI | $0.3 | $0.3 | Serverless |
Frequently Asked Questions
- What is NexusRaven used for?
- NexusRaven is used for structured outputs and coding. The family description and listed model capabilities point to those workloads as the best fit.
- How does NexusRaven compare to Starling?
- NexusRaven by Nexusflow is strongest where you need structured outputs, while Starling by Nexusflow is the closest related family to check for coding. NexusRaven has 2 listed variants and reaches up to 100k context, while Starling reaches up to 8k context, so compare the specs and pricing tables before choosing a production model.
- Which NexusRaven model should I use?
- For the lowest listed input price, start with NexusRaven-V2 13B through Together AI at $0.3/1M input tokens. For the most capable/latest local choice, evaluate NexusRaven-V2 13B with 100k context and structured outputs.
