Pricing
| Type | Price (per 1M) |
|---|---|
| Input tokens | $0.30 |
| Output tokens | $0.30 |
Capabilities
VisionMultimodalReasoningFunction CallingTool UseJSON ModeCode Execution
About NexusRaven-V2 13B
NexusRaven-V2 13B is an open-source, commercially viable large language model crafted by NexusFlow, based on the CodeLlama-13b-Instruct-hf model. It specializes in zero-shot function calling, surpassing GPT-4 by 7% in certain scenarios. This model supports generating single, nested, and parallel function calls and can explain its processes clearly. It performs best with a retriever to manage numerous functions without exceeding its context window, though it can sometimes produce inaccurate function calls, requiring guardrails for error mitigation. With various quantized versions balancing model size and performance, NexusRaven-V2 is also compatible with multiple frameworks like LangChain.
Model Specs
Released2023-10-17
Parameters13B
ArchitectureDecoder Only