WizardLM Team
20 models across 4 families · Latest: WizardLM-2 8x22B (2024-04)
Advancing Language Model Capabilities
WizardLM Team's portfolio covers 19 active models across 4 current families, spanning coding, classification, and json / tool use. Open a model detail page to compare provider routes and sourced benchmarks.
Covers 3 workload areas across 19 active tracked models; last verified 2026-10-13.
Use it for
- Teams evaluating coding, classification, and json / tool use across this lab's releases
- Comparing model families before committing to a flagship
- Migration and pricing follow-ups across 19 tracked models
Do not use it for
- Choosing a hosting provider without opening a model page for price ladders
Active models
19
Current models from this lab, excluding deprecated ones
Active families
4
Current model families from this lab
Open catalog
19 open
9 open source / 10 open weights
Lowest output price
$0.070 /1M
Cheapest tracked output across active models, per 1M tokens
Latest dated release
2024-04-15
WizardLM-2 8x22B
Freshness
2026-10-13
Researched 4d ago
Information
Release cadence
- Dated releases
- 5
- Latest model
- WizardLM-2 8x22B
- Latest date
- 2024-04-15
Where this lab wins
- Coding: 1 tracked model with SWE-bench / HumanEval-style scores.
- Classification: 4 tracked models with MMLU-class moderation/safety coverage.
- JSON/tool-use: 4 tracked models with BFCL / Nexus strict-JSON routing coverage.
Flagship quality / price signal
- Flagship
- WizardLM-2 8x22B
- Selection source
- Best sourced coding Q/$
- Coding grade
- B
- Benchmark / output price
- humaneval 67.5 · $0.500/1M
About
The WizardLM Team is dedicated to advancing large language models (LLMs) with a focus on enhancing their ability to follow complex instructions. Their core innovation lies in the ""AI Align AI"" (AAA) framework, which leverages multiple state-of-the-art LLMs to collaboratively train and refine one another. This approach, combined with techniques like Evol-Instruct and Reinforcement Learning for Instruction and Process Supervision (RLEIF), enables the creation of high-quality training data and the fine-tuning of models for superior performance in complex chat, multilingual capabilities, reasoning, and agent-based interactions.
Featured models
| Model | Released | Context | Input price ($/1M) | Output price ($/1M) | License | Openness |
|---|---|---|---|---|---|---|
| WizardLM-2 8x22B | 2024-04-15 | - | $0.65 | $0.65 | Apache 2.0 | Open source |
| WizardLM-2 70B | 2024-04-15 | - | - | - | Llama 2 Community | Open weights |
| WizardLM-2 7B | 2024-04-15 | - | $0.05 | $0.15 | Apache 2.0 | Open source |
Model families
Recent releases
- WizardLM-2 8x22B- 2024-04-15
- WizardLM-2 70B- 2024-04-15
- WizardLM-2 7B- 2024-04-15
- Wizard Coder 15B- 2024-01-29
- WizardCoder 1B- 2024-01-29
Explore related pages
Last reviewed: 2026-10-17. Data sourced from public lab announcements and provider documentation.