GRPO-Think-14B-4k
Aletheia-Bench/GRPO-Think-14B-4k
Tracked by OpenModelStats since September 4, 2026View on Hugging Face ↗
- Downloads 30DRolling recent-window download count as reported by the source platform.
- 271
- Total downloadsCumulative all-time download counter reported by the source platform.
- 445
- Gained 7DIncrease in cumulative total downloads between OpenModelStats' own observations seven days apart.
- +4(increase)
- Gained 30DIncrease in cumulative total downloads between OpenModelStats' own observations roughly thirty days apart.
- —
- LikesCommunity likes reported by the source platform.
- 0
- +0 in 7D
- Rank · tracked modelsPosition among all eligible models tracked by OpenModelStats, ordered by recent downloads. #11667 among tracked Text Generation models.
- #39,220
- 145(down 145 places)this week
- Parameters
- 14.8B
- Last source update
- Aug 25, 2026
- MomentumOpenModelStats' composite rising-model signal (0–100): a percentile blend of absolute 7-day download gains, capped percentage growth, and likes gained. See Methodology.
- 12.2
- GRPO-Think-14B-4k gained 4 downloads during the last seven tracked days.
- It moved from #11638 to #11667 among tracked Text Generation models this week.
Downloads over time
Cumulative total downloads observed by OpenModelStatsLikes over time
Rank among tracked models
Lower is betterOverview
GRPO-Think-14B-4k is a text generation model published by Aletheia-Bench. OpenModelStats has tracked the model since Sep 4, 2026, most recently observing it Sep 18, 2026.
- Published
- Dec 8, 2025
- Last updated
- Aug 25, 2026
- Library
- transformers
- License
- cc-by-nc-sa-4.0
- Task
- Text Generation
- Parameters
- 14,770,033,664
- Spaces
- 0
- Derivative models
- —
- Velocity 7D/day
- 1
transformerssafetensorsqwen2text-generationcodecode-verificationrlvrgrpoconversationaldataset:Aletheia-Bench/Aletheia-Trainarxiv:2601.12186base_model:deepseek-ai/DeepSeek-R1-Distill-Qwen-14B
Model family
Based on base-model relationships reported by the source's metadata.
Base model
Similar models
| Model | DL 30D |
|---|---|
| Qwen2.5-14B-Instruct-AWQQwen | 1.8M |
| Qwen2.5-Coder-14B-InstructQwen | 1.7M |
| Qwen2.5-14B-InstructQwen | 1.6M |
| JiRackUltra_14bCMSManhattan | 964.2K |
| Qwen2.5-Coder-14B-Instruct-AWQQwen | 427.5K |
| DeepSeek-R1-Distill-Qwen-14Bdeepseek-ai | 337.3K |