Model reference · open weights
CapRL-Qwen3VL is an open-weight language model from internlm. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.
Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.
What it is
| Maker | internlm |
|---|---|
| Type | Language models |
| Task | Vision + text |
| Parameters (lead) | 4.4B |
| Runs with | transformers |
| Released | 2025-12-24 |
| Popularity | 3k downloads / month |
| Licence | Open weights |
About
📖Paper | 🏠Github | 🤗CapRL Collection | 🤗Daily Paper
| Series | Models & Resources |
|---|---|
| CapRL 2.0 Series | 🤗 CapRL-Qwen3VL-2B | 🤗 CapRL-Qwen3VL-4B | 📦 CapRL-Qwen3VL-2B-GGUF | 📦 CapRL-Qwen3VL-4B-GGUF | 🌈CapRL-Qwen3VL-4B Space |
| CapRL 1.0 Series | 🤗 CapRL-Qwen2.5VL-3B | 🤗 CapRL-InternVL3.5-8B |📊 CapRL-QA-75K Dataset | 📊 CapRL-2M Dataset | 📦 CapRL-3B-GGUF | 📦 CapRL-3B-i1-GGUF | 🌈CapRL-Qwen2.5VL-3B Space |
We are excited to release the CapRL 2.0 series: CapRL-Qwen3VL-2B and CapRL-Qwen3VL-4B. These models feature fewer parameters while delivering even more powerful captioning performance. Notably, CapRL-Qwen3VL-4B significantly outperforms both CapRL-Qwen2.5VL-3B and Qwen2.5VL-72B in captioning tasks, establishing itself as the top-performing model for captioning within the CapRL series. This leap in efficiency is driven by our upgraded training recipe, which includes a more rigorous QA data filter and a significantly more diverse image dataset. We welcome everyone to try them out!
When selecting between the available CapRL models, it's essential to consider the trade-off between performance and computational cost. This guide will help you choose the most suitable model for your specific needs:
| Model | Parameters | Strength |
|---|---|---|
| 🤗CapRL-Qwen3VL-2B | 2B | Speed, Efficiency |
| 🤗CapRL-Qwen3VL-4B | 4B | High Performance, Advanced Captioning Ability |
Now you can try out CapRL with your own images🎨! ➡️ 🌈CapRL-Qwen2.5VL-3B Space and 🌈CapRL-Qwen3VL-4B Space.
We are working on even stronger base models and upgrading our training recipe — stay tuned!
We are excited to introduce CapRL-3B, a lightweight 3B image captioner that achieves perception capabilities comparable to Qwen2.5-VL-72B.
This is the first study of applying Reinforcement Learning with Verifiable Rewards for the open-ended and subjective image captioning task. Unlike traditional Supervised Fine-Tuning, which can lead to models memorizing a limited set of annotated captions, our method allows the model to explore and generate a broader range of creative and general descriptions. CapRL is a new training paradigm featuring a decoupled two-stage pipeline. The initial stage uses LVLMs to generate rich and accurate captions. Subsequently, the second stage evaluates caption quality by using a vision-only LLM to perform the QA task. We also created a specific QA curation pipeline to ensure the quality of the questions and answers used for the second stage.
By employing the CapRL training framework, initializing with the Qwen2.5-VL-3B model, and using a carefully filtered 75K QA dataset as the training set, we obtained a highly capable captioner, CapRL-3B.
If you want to use CapRL-3B for captioning, you can di
From the published model card. Full card on the HuggingFace links in the sidebar.
Using it via the API
Once AxForge deploys caprl-qwen3vl for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (caprl-qwen3vl below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/chat/completions \
-H "Authorization: Bearer $AXFORGE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"caprl-qwen3vl","messages":[{"role":"user","content":"Hello"}]}'
Create an account — your API key is available in the console. 5M tokens/month currently included with every new account at launch.