WizardLM 70B v1
WizardLM 70B v1 is WizardLM's flagship instruction-following model, built by fine-tuning Llama 2 70B using the Evol-Instruct data augmentation technique. Evol-Instruct systematically rewrites training instructions to span simple to highly complex prompts, which gives the resulting model an unusually robust ability to follow nuanced, layered, and difficult instructions.
At 70B scale it combines the broad knowledge of the Llama 2 base with WizardLM's strong instruction-following alignment. It performs well on open-ended reasoning, detailed analytical questions, and task decomposition. Teams looking for a powerful open-weight instruction model for enterprise-style assistants, research pipelines, or document processing workflows find WizardLM 70B a solid self-hostable option.
Key Features
Strong complex instruction-following via Evol-Instruct training
Detailed analytical reasoning and multi-step task decomposition
Long-form writing with consistent tone and structure
Question answering with nuanced, multi-part responses
Document summarization and structured data extraction
Open-weight model suitable for on-premise enterprise deployment
Ideal Use Cases
Enterprise instruction-following assistant deployed on-premise
Research pipeline requiring detailed analytical output
Document processing and structured summarization
Complex question-answering for knowledge management tools
Baseline for Llama 2 70B instruction-following comparisons
Example Prompts for WizardLM 70B v1
Technical Specifications
| Provider | WizardLM |
| Category | Text |
| Modality | Text -> Text |
Frequently Asked Questions
Try WizardLM 70B v1 now
Start using WizardLM 70B v1 instantly — 100 free credits, no credit card required. Access 750+ AI models through one platform.