LLM-jp-4.1 33B Thinking
LLM-jp-4.1 33B Thinking updates post-training for the 33B architecture. It retains a 65,536-token context and publishes the SFT and model-specific DPO datasets.
Research release
The revision changes post-training
The card specifies SFT and DPO without reinforcement learning. It records 11.7 trillion pretraining and mid-training tokens. Its chat template supports Harmony-style messages, but requires the included tokenizer. Some pretraining material remains unpublished because of licensing limits.LLM-jp-4.1 33B Thinking official model card
Inspect task gains separately
Practical guidance. Review reasoning, instruction following and tool use separately. A single average score can hide changes in the tasks that matter to an application.
GGUF is a distribution of this model
The project also publishes GGUF files. Its reviewed card says llama.cpp needs the project's tokenizer-handling fork; an unmodified upstream build can fail to parse chat messages. These files remain part of the same trained release.LLM-jp-4.1 33B Thinking official GGUF release
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Launch date | |
| Access | Downloadable |
| Model identifiers | llm-jp-4.1-33b-thinkingllm-jp-4.1-33b-thinking-gguf |
| Context window | 65,536 tokens Maximum sequence length stated in the model card. LLM-jp-4.1 33B Thinking official model card |
| Parameters | 33,219,548,160 total parameters.LLM-jp-4.1 33B Thinking official model card |
| Architecture | Dense; 64 layers, hidden size 5,120.LLM-jp-4.1 33B Thinking official model card |
| License | Apache 2.0; research weights, not a provider-operated API.LLM-jp-4.1 33B Thinking official model card |
Related models
- LLM-jp-4 33B Thinking
LLM-jp / NII LLMC ยท LLM-jp-4
LLM-jp-4 33B Thinking provides Japanese and English text reasoning weights with a 65,536-token context.
Launch date not confirmedResearch releaseDownloadable