Hermes 3 70B
Hermes 3 70B is Nous Research's third-generation Hermes fine-tune, applied to Meta's Llama 3 70B base model. The Hermes series is known for improving instruction following, tool use, and structured output quality over the base model, and the 70B scale brings substantial capacity for reasoning-heavy tasks.
The model is a strong performer for agentic workflows, function-calling applications, and complex reasoning chains. Nous Research's fine-tuning philosophy emphasizes helpfulness and coherence over default conservatism, resulting in a model that produces substantive, direct responses across coding, analysis, and knowledge-intensive tasks.
Key Features
Fine-tuned on Llama 3 70B for improved instruction adherence
Strong function calling and structured JSON output support
Enhanced multi-step reasoning and chain-of-thought capability
Well-suited to agentic tool-use pipelines
Direct, substantive response style with reduced over-refusal
Broad knowledge coverage from the Llama 3 pretraining corpus
Ideal Use Cases
Agentic applications requiring reliable tool and function calling
Complex question answering and research assistance
Code generation and technical problem solving
Structured data extraction and JSON output generation
Multi-step planning and reasoning tasks
Example Prompts for Hermes 3 70B
Technical Specifications
| Provider | NousResearch |
| Category | Text |
| Modality | Text -> Text |
Frequently Asked Questions
Try Hermes 3 70B now
Start using Hermes 3 70B instantly — 100 free credits, no credit card required. Access 750+ AI models through one platform.