Key Specifications
| Vendor | other |
|---|
| Version | chatglm3-6b |
|---|
| Release Date | 2023-10-27 |
|---|
| Context Window | 32768 tokens |
|---|
| Input Modalities | text |
|---|
| Output Modalities | text |
|---|
| License | GLM-3 License |
|---|
| Documentation | https://huggingface.co/models |
|---|
Benchmark Performance
| Benchmark | Score | Unit | Evaluated At | Notes | Source |
|---|
| MMLU | 58 | % | 2023-10-27 | 5-shot | view |
| HUMANEVAL | 51 | pass@1 | 2023-10-27 | — | view |
| GSM8K | 37.9 | % | 2023-10-27 | 0-shot CoT | view |
| MATH | 31.8 | % | 2023-10-27 | 0-shot CoT | view |
| BBH | 55.7 | % | 2023-10-27 | 3-shot CoT | view |
| GPQA | 20.1 | % | 2023-10-27 | 0-shot | view |
| IFEVAL | 42.5 | % | 2023-10-27 | prompt_strict | view |
| ARC | 88.9 | % | 2023-10-27 | challenge | view |
| MUSR | 42.9 | % | 2023-10-27 | 0-shot | view |
| WINOGRANDE | 67.7 | % | 2023-10-27 | 0-shot | view |
Pricing
| Tier | Price | Currency |
|---|
| Input | $0.18 / Mtok | USD |
| Output | $0.18 / Mtok | USD |
| Cache Read | $0 / Mtok | USD |
| Cache Write | $0 / Mtok | USD |
Source:
https://huggingface.co/models
· as of 2023-10-27
Compliance
- Data Residency: self-host
- SOC2: ✗
- HIPAA: ✗
- GDPR: ✗
- ISO 27001: ✗
ChatGLM3 6B
Model Overview
清华智谱 ChatGLM3 6B 第三代开源模型, 32K 上下文, 6B 参数, 中英双语对话能力突出, 适合本地部署。
Core Specifications
| Vendor | Version | Release Date | Context Window | Input Modalities | Output Modalities | License |
|---|
| Other | chatglm3-6b | 2023-10-27 | 32K | text | text | GLM-3 License |
| Benchmark | Score | Unit | Notes |
|---|
| MMLU (Massive Multitask Language Understanding) | 58.0 | % | 5-shot |
| HumanEval | 51.0 | pass@1 | — |
| GSM8K (Grade School Math 8K) | 37.9 | % | 0-shot CoT |
| MATH | 31.8 | % | 0-shot CoT |
| BBH (BIG-Bench Hard) | 55.7 | % | 3-shot CoT |
| GPQA | 20.1 | % | 0-shot |
| IFEval | 42.5 | % | prompt_strict |
| ARC | 88.9 | % | challenge |
| MUSR | 42.9 | % | 0-shot |
| WinoGrande | 67.7 | % | 0-shot |
Pricing
| Input | Output | Cache Read | Cache Write |
|---|
| — | — | — | — |
per million tokens
Strengths
- Reliable general-purpose model.
Weaknesses
- MMLU 58.0, weak knowledge reasoning.
- Proprietary, not self-hostable.
Use Cases
References