LLM-jp-3 172B Base
LLM-jp-3 172B Base completes the published pretraining sequence at 2.1 trillion tokens. Its weights require individual permission under the model's terms.
Restricted access
Service use and weight redistribution have different terms
The terms allow services built with the approved model, but do not allow service users to obtain its weights directly. This access arrangement differs from the Apache-licensed smaller dense models and the later Instruct3 license.LLM-jp-3 172B Base official model card
Check the intended distribution method
Practical guidance. The release can support approved internal research or service use. It is a poor fit for a project whose delivery requires customers to receive modified model files.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Restricted access |
| Access | Downloadable |
| Model identifiers | llm-jp-3-172b |
| Context window | 4,096 tokens Maximum sequence length stated in the model card. LLM-jp-3 172B Base official model card |
| Architecture | 96 layers; hidden size 12,288; 4,096-token context.LLM-jp-3 172B Base official model card |
| Parameters | 172,392,173,568 total, from embedding and non-embedding counts.LLM-jp-3 172B Base official model card |
| License | Custom terms permit commercial and noncommercial use after approval, but prohibit redistribution of weights and modified weights.LLM-jp-3 172B Base official model card |
Related models
- LLM-jp-3 172B beta2
LLM-jp / NII LLMC · LLM-jp-3 172B
LLM-jp-3 172B beta2 is a gated 172B research checkpoint with 1.4 trillion pretraining tokens.
Launch date not confirmedRestricted accessDownloadable - LLM-jp-3 172B Instruct2
LLM-jp / NII LLMC · LLM-jp-3 172B
A gated 172B supervised instruction release for comparison with preference-aligned Instruct3.
Launch date not confirmedRestricted accessDownloadable - LLM-jp-3 172B Instruct3
LLM-jp / NII LLMC · LLM-jp-3 172B
The 172B preference-aligned release has a separate license that permits redistribution subject to attribution and other terms.
Launch date not confirmedResearch releaseDownloadable