List of Supported Models#
vLLM supports generative and pooling models across various tasks.
If a model supports more than one task, you can set the task via the --task argument.
For each task, we list the model architectures that have been implemented in vLLM. Alongside each architecture, we include some popular models that use it.
Loading a Model#
HuggingFace Hub#
By default, vLLM loads models from HuggingFace (HF) Hub.
To determine whether a given model is natively supported, you can check the config.json file inside the HF repository.
If the "architectures" field contains a model architecture listed below, then it should be natively supported.
Models do not need to be natively supported to be used in vLLM. The Transformers fallback enables you to run models directly using their Transformers implementation (or even remote code on the Hugging Face Model Hub!).
Tip
The easiest way to check if your model is really supported at runtime is to run the program below:
from vllm import LLM
# For generative models (task=generate) only
llm = LLM(model=..., task="generate") # Name or path of your model
output = llm.generate("Hello, my name is")
print(output)
# For pooling models (task={embed,classify,reward,score}) only
llm = LLM(model=..., task="embed") # Name or path of your model
output = llm.encode("Hello, my name is")
print(output)
If vLLM successfully returns text (for generative models) or hidden states (for pooling models), it indicates that your model is supported.
Otherwise, please refer to Adding a New Model for instructions on how to implement your model in vLLM. Alternatively, you can open an issue on GitHub to request vLLM support.
Transformers fallback#
vLLM can fallback to model implementations that are available in Transformers. This does not work for all models for now, but most decoder language models are supported, and vision language model support is planned!
To check if the backend is Transformers, you can simply do this:
from vllm import LLM
llm = LLM(model=..., task="generate") # Name or path of your model
llm.apply_model(lambda model: print(type(model)))
If it is TransformersModel then it means it’s based on Transformers!
Tip
You can force the use of TransformersModel by setting model_impl="transformers" for Offline Inference or --model-impl transformers for the OpenAI-Compatible Server.
Note
vLLM may not fully optimise the Transformers implementation so you may see degraded performance if comparing a native model to a Transformers model in vLLM.
Supported features#
The Transformers fallback explicitly supports the following features:
Quantization (except GGUF)
Distributed Inference and Serving (pipeline parallel coming soon Pull Request #12832!)
Remote code#
Earlier we mentioned that the Transformers fallback enables you to run remote code models directly in vLLM. If you are interested in this feature, this section is for you!
Simply set trust_remote_code=True and vLLM will run any model on the Model Hub that is compatible with Transformers.
Provided that the model writer implements their model in a compatible way, this means that you can run new models before they are officially supported in Transformers or vLLM!
from vllm import LLM
llm = LLM(model=..., task="generate", trust_remote_code=True) # Name or path of your model
llm.apply_model(lambda model: print(model.__class__))
To make your model compatible with the Transformers fallback, it needs:
from transformers import PreTrainedModel
from torch import nn
class MyAttention(nn.Module):
def forward(self, hidden_states, **kwargs): # <- kwargs are required
...
attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]
attn_output, attn_weights = attention_interface(
self,
query_states,
key_states,
value_states,
**kwargs,
)
...
class MyModel(PreTrainedModel):
_supports_attention_backend = True
Here is what happens in the background:
The config is loaded
MyModelPython class is loaded from theauto_map, and we check that the model_supports_attention_backend.The
TransformersModelbackend is used. See vllm/model_executor/models/transformers.py, which leverageself.config._attn_implementation = "vllm", thus the need to useALL_ATTENTION_FUNCTION.
To make your model compatible with tensor parallel, it needs:
from transformers import PretrainedConfig
class MyConfig(PretrainedConfig):
base_model_tp_plan = {
"layers.*.self_attn.q_proj": "colwise",
...
}
Tip
base_model_tp_plan is a dict that maps fully qualified layer name patterns to tensor parallel styles (currently only "colwise" and "rowwise" are supported).
That’s it!
ModelScope#
To use models from ModelScope instead of HuggingFace Hub, set an environment variable:
export VLLM_USE_MODELSCOPE=True
And use with trust_remote_code=True.
from vllm import LLM
llm = LLM(model=..., revision=..., task=..., trust_remote_code=True)
# For generative models (task=generate) only
output = llm.generate("Hello, my name is")
print(output)
# For pooling models (task={embed,classify,reward,score}) only
output = llm.encode("Hello, my name is")
print(output)
List of Text-only Language Models#
Generative Models#
See this page for more information on how to use generative models.
Text Generation (--task generate)#
Architecture |
Models |
Example HF Models |
||
|---|---|---|---|---|
|
Aquila, Aquila2 |
|
✅︎ |
✅︎ |
|
Arctic |
|
✅︎ |
|
|
Baichuan2, Baichuan |
|
✅︎ |
✅︎ |
|
BLOOM, BLOOMZ, BLOOMChat |
|
✅︎ |
|
|
BART |
|
||
|
ChatGLM |
|
✅︎ |
✅︎ |
|
Command-R |
|
✅︎ |
✅︎ |
|
DBRX |
|
✅︎ |
|
|
DeciLM |
|
✅︎ |
|
|
DeepSeek |
|
✅︎ |
|
|
DeepSeek-V2 |
|
✅︎ |
|
|
DeepSeek-V3 |
|
✅︎ |
|
|
EXAONE-3 |
|
✅︎ |
✅︎ |
|
Falcon |
|
✅︎ |
|
|
FalconMamba |
|
✅︎ |
✅︎ |
|
Gemma |
|
✅︎ |
✅︎ |
|
Gemma 2 |
|
✅︎ |
✅︎ |
|
Gemma 3 |
|
✅︎ |
✅︎ |
|
GLM-4 |
|
✅︎ |
✅︎ |
|
GPT-2 |
|
✅︎ |
|
|
StarCoder, SantaCoder, WizardCoder |
|
✅︎ |
✅︎ |
|
GPT-J |
|
✅︎ |
|
|
GPT-NeoX, Pythia, OpenAssistant, Dolly V2, StableLM |
|
✅︎ |
|
|
Granite 3.0, Granite 3.1, PowerLM |
|
✅︎ |
✅︎ |
|
Granite 3.0 MoE, PowerMoE |
|
✅︎ |
✅︎ |
|
Granite MoE Shared |
|
✅︎ |
✅︎ |
|
GritLM |
|
✅︎ |
✅︎ |
|
Grok1 |
|
✅︎ |
✅︎ |
|
InternLM |
|
✅︎ |
✅︎ |
|
InternLM2 |
|
✅︎ |
✅︎ |
|
InternLM3 |
|
✅︎ |
✅︎ |
|
Jais |
|
✅︎ |
|
|
Jamba |
|
✅︎ |
✅︎ |
|
Llama 3.1, Llama 3, Llama 2, LLaMA, Yi |
|
✅︎ |
✅︎ |
|
Mamba |
|
✅︎ |
|
|
MiniCPM |
|
✅︎ |
✅︎ |
|
MiniCPM3 |
|
✅︎ |
✅︎ |
|
Mistral, Mistral-Instruct |
|
✅︎ |
✅︎ |
|
Mixtral-8x7B, Mixtral-8x7B-Instruct |
|
✅︎ |
✅︎ |
|
MPT, MPT-Instruct, MPT-Chat, MPT-StoryWriter |
|
✅︎ |
|
|
Nemotron-3, Nemotron-4, Minitron |
|
✅︎ |
✅︎ |
|
OLMo |
|
✅︎ |
|
|
OLMo2 |
|
✅︎ |
|
|
OLMoE |
|
✅︎ |
✅︎ |
|
OPT, OPT-IML |
|
✅︎ |
|
|
Orion |
|
✅︎ |
|
|
Phi |
|
✅︎ |
✅︎ |
|
Phi-4, Phi-3 |
|
✅︎ |
✅︎ |
|
Phi-3-Small |
|
✅︎ |
|
|
Phi-3.5-MoE |
|
✅︎ |
✅︎ |
|
Persimmon |
|
✅︎ |
|
|
Qwen |
|
✅︎ |
✅︎ |
|
QwQ, Qwen2 |
|
✅︎ |
✅︎ |
|
Qwen2MoE |
|
✅︎ |
|
|
StableLM |
|
✅︎ |
|
|
Starcoder2 |
|
✅︎ |
|
|
Solar Pro |
|
✅︎ |
✅︎ |
|
TeleChat2 |
|
✅︎ |
✅︎ |
|
TeleFLM |
|
✅︎ |
✅︎ |
|
XVERSE |
|
✅︎ |
✅︎ |
|
Zamba2 |
|
Note
Currently, the ROCm version of vLLM supports Mistral and Mixtral only for context lengths up to 4096.
Pooling Models#
See this page for more information on how to use pooling models.
Important
Since some model architectures support both generative and pooling tasks, you should explicitly specify the task type to ensure that the model is used in pooling mode instead of generative mode.
Text Embedding (--task embed)#
Architecture |
Models |
Example HF Models |
||
|---|---|---|---|---|
|
BERT-based |
|
||
|
Gemma 2-based |
|
✅︎ |
|
|
GritLM |
|
✅︎ |
✅︎ |
|
Llama-based |
|
✅︎ |
✅︎ |
|
Qwen2-based |
|
✅︎ |
✅︎ |
|
RoBERTa-based |
|
||
|
XLM-RoBERTa-based |
|
Note
ssmits/Qwen2-7B-Instruct-embed-base has an improperly defined Sentence Transformers config.
You should manually set mean pooling by passing --override-pooler-config '{"pooling_type": "MEAN"}'.
Note
The HF implementation of Alibaba-NLP/gte-Qwen2-1.5B-instruct is hardcoded to use causal attention despite what is shown in config.json. To compare vLLM vs HF results,
you should set --hf-overrides '{"is_causal": true}' in vLLM so that the two implementations are consistent with each other.
For both the 1.5B and 7B variants, you also need to enable --trust-remote-code for the correct tokenizer to be loaded.
See relevant issue on HF Transformers.
If your model is not in the above list, we will try to automatically convert the model using
as_embedding_model(). By default, the embeddings
of the whole prompt are extracted from the normalized hidden state corresponding to the last token.
Reward Modeling (--task reward)#
Architecture |
Models |
Example HF Models |
||
|---|---|---|---|---|
|
InternLM2-based |
|
✅︎ |
✅︎ |
|
Llama-based |
|
✅︎ |
✅︎ |
|
Qwen2-based |
|
✅︎ |
✅︎ |
|
Qwen2-based |
|
✅︎ |
✅︎ |
If your model is not in the above list, we will try to automatically convert the model using
as_reward_model(). By default, we return the hidden states of each token directly.
Important
For process-supervised reward models such as peiyi9979/math-shepherd-mistral-7b-prm, the pooling config should be set explicitly,
e.g.: --override-pooler-config '{"pooling_type": "STEP", "step_tag_id": 123, "returned_token_ids": [456, 789]}'.
Classification (--task classify)#
Architecture |
Models |
Example HF Models |
||
|---|---|---|---|---|
|
Jamba |
|
✅︎ |
✅︎ |
|
Qwen2-based |
|
✅︎ |
✅︎ |
If your model is not in the above list, we will try to automatically convert the model using
as_classification_model(). By default, the class probabilities are extracted from the softmaxed hidden state corresponding to the last token.
Sentence Pair Scoring (--task score)#
List of Multimodal Language Models#
The following modalities are supported depending on the model:
Text
Image
Video
Audio
Any combination of modalities joined by + are supported.
e.g.:
T + Imeans that the model supports text-only, image-only, and text-with-image inputs.
On the other hand, modalities separated by / are mutually exclusive.
e.g.:
T / Imeans that the model supports text-only and image-only inputs, but not text-with-image inputs.
See this page on how to pass multi-modal inputs to the model.
Important
To enable multiple multi-modal items per text prompt, you have to set limit_mm_per_prompt (offline inference)
or --limit-mm-per-prompt (online serving). For example, to enable passing up to 4 images per text prompt:
Offline inference:
llm = LLM(
model="Qwen/Qwen2-VL-7B-Instruct",
limit_mm_per_prompt={"image": 4},
)
Online serving:
vllm serve Qwen/Qwen2-VL-7B-Instruct --limit-mm-per-prompt image=4
Note
vLLM currently only supports adding LoRA to the language backbone of multimodal models.
Generative Models#
See this page for more information on how to use generative models.
Text Generation (--task generate)#
Architecture |
Models |
Inputs |
Example HF Models |
|||
|---|---|---|---|---|---|---|
|
Aria |
T + I+ |
|
✅︎ |
✅︎ |
|
|
BLIP-2 |
T + IE |
|
✅︎ |
✅︎ |
|
|
Chameleon |
T + I |
|
✅︎ |
✅︎ |
|
|
DeepSeek-VL2 |
T + I+ |
|
✅︎ |
✅︎ |
|
|
Florence-2 |
T + I |
|
|||
|
Fuyu |
T + I |
|
✅︎ |
✅︎ |
|
|
Gemma 3 |
T + I+ |
|
✅︎ |
✅︎ |
⚠️ |
|
GLM-4V |
T + I |
|
✅︎ |
✅︎ |
✅︎ |
|
H2OVL |
T + IE+ |
|
✅︎ |
✅︎* |
|
|
Idefics3 |
T + I |
|
✅︎ |
✅︎ |
|
|
InternVideo 2.5, InternVL 2.5, Mono-InternVL, InternVL 2.0 |
T + IE+ |
|
✅︎ |
✅︎ |
|
|
LLaVA-1.5 |
T + IE+ |
|
✅︎ |
✅︎ |
|
|
LLaVA-NeXT |
T + IE+ |
|
✅︎ |
✅︎ |
|
|
LLaVA-NeXT-Video |
T + V |
|
✅︎ |
✅︎ |
|
|
LLaVA-Onevision |
T + I+ + V+ |
|
✅︎ |
✅︎ |
|
|
MiniCPM-O |
T + IE+ + VE+ + AE+ |
|
✅︎ |
✅︎ |
|
|
MiniCPM-V |
T + IE+ + VE+ |
|
✅︎ |
✅︎ |
|
|
Llama 3.2 |
T + I+ |
|
|||
|
Molmo |
T + I |
|
✅︎ |
✅︎ |
✅︎ |
|
NVLM-D 1.0 |
T + I+ |
|
✅︎ |
✅︎ |
|
|
PaliGemma, PaliGemma 2 |
T + IE |
|
✅︎ |
⚠️ |
|
|
Phi-3-Vision, Phi-3.5-Vision |
T + IE+ |
|
✅︎ |
✅︎ |
|
|
Phi-4-multimodal |
T + I+ / T + A+ / I+ + A+ |
|
✅︎ |
||
|
Pixtral |
T + I+ |
|
✅︎ |
✅︎ |
|
|
Qwen-VL |
T + IE+ |
|
✅︎ |
✅︎ |
✅︎ |
|
Qwen2-Audio |
T + A+ |
|
✅︎ |
✅︎ |
|
|
QVQ, Qwen2-VL |
T + IE+ + VE+ |
|
✅︎ |
✅︎ |
✅︎ |
|
Qwen2.5-VL |
T + IE+ + VE+ |
|
✅︎ |
✅︎ |
✅︎ |
|
Ultravox |
T + AE+ |
|
✅︎ |
✅︎ |
✅︎ |
^ You need to set the architecture name via --hf-overrides to match the one in vLLM.
• For example, to use DeepSeek-VL2 series models:
--hf-overrides '{"architectures": ["DeepseekVLV2ForCausalLM"]}'
E Pre-computed embeddings can be inputted for this modality.
+ Multiple items can be inputted per text prompt for this modality.
Important
To use Gemma3 series models, you have to install Hugging Face Transformers library from source via
pip install git+https://github.com/huggingface/transformers.
Pan-and-scan image pre-processing is currently supported on V0 (but not V1).
You can enable it by passing --mm-processor-kwargs '{"do_pan_and_scan": True}'.
Warning
Both V0 and V1 support Gemma3ForConditionalGeneration for text-only inputs.
However, there are differences in how they handle text + image inputs:
V0 correctly implements the model’s attention pattern:
Uses bidirectional attention between the image tokens corresponding to the same image
Uses causal attention for other tokens
Implemented via (naive) PyTorch SDPA with masking tensors
Note: May use significant memory for long prompts with image
V1 currently uses a simplified attention pattern:
Uses causal attention for all tokens, including image tokens
Generates reasonable outputs but does not match the original model’s attention for text + image inputs, especially when
{"do_pan_and_scan": True}Will be updated in the future to support the correct behavior
This limitation exists because the model’s mixed attention pattern (bidirectional for images, causal otherwise) is not yet supported by vLLM’s attention backends.
Note
h2oai/h2ovl-mississippi-2b will be available in V1 once we support backends other than FlashAttention.
Note
To use TIGER-Lab/Mantis-8B-siglip-llama3, you have to pass --hf_overrides '{"architectures": ["MantisForConditionalGeneration"]}' when running vLLM.
Note
The official openbmb/MiniCPM-V-2 doesn’t work yet, so we need to use a fork (HwwwH/MiniCPM-V-2) for now.
For more details, please see: Pull Request #4087
Warning
Our PaliGemma implementations have the same problem as Gemma 3 (see above) for both V0 and V1.
Pooling Models#
See this page for more information on how to use pooling models.
Important
Since some model architectures support both generative and pooling tasks, you should explicitly specify the task type to ensure that the model is used in pooling mode instead of generative mode.
Text Embedding (--task embed)#
Any text generation model can be converted into an embedding model by passing --task embed.
Note
To get the best results, you should use pooling models that are specifically trained as such.
The following table lists those that are tested in vLLM.
Transcription (--task transcription)#
Speech2Text models trained specifically for Automatic Speech Recognition.
Model Support Policy#
At vLLM, we are committed to facilitating the integration and support of third-party models within our ecosystem. Our approach is designed to balance the need for robustness and the practical limitations of supporting a wide range of models. Here’s how we manage third-party model support:
Community-Driven Support: We encourage community contributions for adding new models. When a user requests support for a new model, we welcome pull requests (PRs) from the community. These contributions are evaluated primarily on the sensibility of the output they generate, rather than strict consistency with existing implementations such as those in transformers. Call for contribution: PRs coming directly from model vendors are greatly appreciated!
Best-Effort Consistency: While we aim to maintain a level of consistency between the models implemented in vLLM and other frameworks like transformers, complete alignment is not always feasible. Factors like acceleration techniques and the use of low-precision computations can introduce discrepancies. Our commitment is to ensure that the implemented models are functional and produce sensible results.
Tip
When comparing the output of
model.generatefrom HuggingFace Transformers with the output ofllm.generatefrom vLLM, note that the former reads the model’s generation config file (i.e., generation_config.json) and applies the default parameters for generation, while the latter only uses the parameters passed to the function. Ensure all sampling parameters are identical when comparing outputs.Issue Resolution and Model Updates: Users are encouraged to report any bugs or issues they encounter with third-party models. Proposed fixes should be submitted via PRs, with a clear explanation of the problem and the rationale behind the proposed solution. If a fix for one model impacts another, we rely on the community to highlight and address these cross-model dependencies. Note: for bugfix PRs, it is good etiquette to inform the original author to seek their feedback.
Monitoring and Updates: Users interested in specific models should monitor the commit history for those models (e.g., by tracking changes in the main/vllm/model_executor/models directory). This proactive approach helps users stay informed about updates and changes that may affect the models they use.
Selective Focus: Our resources are primarily directed towards models with significant user interest and impact. Models that are less frequently used may receive less attention, and we rely on the community to play a more active role in their upkeep and improvement.
Through this approach, vLLM fosters a collaborative environment where both the core development team and the broader community contribute to the robustness and diversity of the third-party models supported in our ecosystem.
Note that, as an inference engine, vLLM does not introduce new models. Therefore, all models supported by vLLM are third-party models in this regard.
We have the following levels of testing for models:
Strict Consistency: We compare the output of the model with the output of the model in the HuggingFace Transformers library under greedy decoding. This is the most stringent test. Please refer to models tests for the models that have passed this test.
Output Sensibility: We check if the output of the model is sensible and coherent, by measuring the perplexity of the output and checking for any obvious errors. This is a less stringent test.
Runtime Functionality: We check if the model can be loaded and run without errors. This is the least stringent test. Please refer to functionality tests and examples for the models that have passed this test.
Community Feedback: We rely on the community to provide feedback on the models. If a model is broken or not working as expected, we encourage users to raise issues to report it or open pull requests to fix it. The rest of the models fall under this category.