# LMDeploy vs vLLM

**Leader by vioscaleAI score:** vLLM

| Attribute | LMDeploy | vLLM |
|---|---|---|
| **vioscaleAI score** | 55.9 (64% (medium)) | 63.5 (58% (medium)) |
| activity.commits_last_30d | 66 | 100 |
| adoption.dependent_repos | 2 | 5 |
| adoption.github_stars | 8,104 | 93,797 |
| adoption.package_downloads_weekly | - | 463,948 |
| content.faq | `[{"answer":"A software framework that enables developers to compress, deploy, and serve large language models with quantization optimization, multiple inference engines, and compatibility across various model architectures. It is indexed under Model Serving.","source":"https://lmdeploy.readthedocs.io/en/latest/","question":"What is LMDeploy?","confidence":0.6},{"answer":"LMDeploy is open source, so it can be self-hosted and used at no licence cost. It is released under the Apache-2.0 licence. Pricing changes often, so verify at source before relying on it.","source":"https://lmdeploy.readthedocs.io/en/latest/","question":"Is LMDeploy free to use?","confidence":0.6},{"answer":"LMDeploy supports Windows and a command-line interface. Platforms we have not confirmed are simply not listed here rather than ruled out.","source":"https://lmdeploy.readthedocs.io/en/latest/","question":"What platforms does LMDeploy support?","confidence":0.6},{"answer":"Yes. LMDeploy can be deployed self-hosted, so it does not have to run on the vendor's infrastructure.","source":"https://lmdeploy.readthedocs.io/en/latest/","question":"Can LMDeploy be self-hosted?","confidence":0.6},{"answer":"Yes. LMDeploy is published under the Apache-2.0 licence, a permissive licence that generally allows commercial use and modification. Licence terms can change between releases, so verify against the repository for the version you intend to use.","source":"https://github.com/internlm/lmdeploy","question":"Is LMDeploy open source?","confidence":0.95}]` | `[{"answer":"An open-source framework that provides optimized LLM inference with low latency and high throughput. It includes continuous batching, memory-efficient attention mechanisms, quantization support, and distributed serving across diverse hardware platforms. It is indexed under MLOps & LLMOps Tools.","source":"https://docs.vllm.ai","question":"What is vLLM?","confidence":0.6},{"answer":"vLLM is open source, so it can be self-hosted and used at no licence cost. It is released under the Apache-2.0 licence. Pricing changes often, so verify at source before relying on it.","source":"https://docs.vllm.ai","question":"Is vLLM free to use?","confidence":0.6},{"answer":"vLLM supports Linux and a command-line interface. Platforms we have not confirmed are simply not listed here rather than ruled out.","source":"https://docs.vllm.ai","question":"What platforms does vLLM support?","confidence":0.6},{"answer":"Yes. vLLM can be deployed cloud / SaaS, on-premise and self-hosted, so it does not have to run on the vendor's infrastructure.","source":"https://docs.vllm.ai","question":"Can vLLM be self-hosted?","confidence":0.6},{"answer":"We have confirmed 32 integrations for vLLM, including Hugging Face, NVIDIA Dynamo, OpenAI-compatible API, Anthropic Messages API, FlashAttention, FlashInfer, CUTLASS and torch.compile, plus 24 more. This is what we could verify from public sources, so the vendor may support others we have not indexed.","source":"https://docs.vllm.ai","question":"What does vLLM integrate with?","confidence":0.6},{"answer":"Yes. vLLM is published under the Apache-2.0 licence, a permissive licence that generally allows commercial use and modification. Licence terms can change between releases, so verify against the repository for the version you intend to use.","source":"https://github.com/vllm-project/vllm","question":"Is vLLM open source?","confidence":0.95}]` |
| deployment.options | `{"self_hosted":true}` | `{"cloud":true,"on_prem":true,"self_hosted":true}` |
| description.long | LMDeploy is a comprehensive toolkit enabling efficient compression, deployment, and serving of large language models across multiple machines and GPUs with advanced optimization techniques. | A production-grade inference engine designed for efficient serving of large language models with support for distributed deployment, GPU acceleration, and multi-model endpoints. |
| features.capabilities | `{"gpu_acceleration":true,"continuous_batching":true,"multi_model_serving":true,"multi_gpu_multi_node":true,"quantization_support":true}` | `{"role":"serving","open_source":true,"managed_cloud":false,"model_serving":true,"self_hostable":true,"multi_provider":true,"vpc_deployment":true,"otel_compatible":true,"gpu_acceleration":true,"kubernetes_native":true,"llm_observability":true,"framework_agnostic":true,"continuous_batching":true,"multi_model_serving":true,"multi_gpu_multi_node":true,"quantization_support":true,"openai_compatible_api":true,"multi_framework_support":true}` |
| integrations.count | 2 | 29 |
| integrations.list | `[{"name":"llm-compressor"},{"name":"OpenCompass"}]` | `[{"name":"Hugging Face"},{"name":"NVIDIA Dynamo"},{"name":"OpenAI-compatible API"},{"name":"Anthropic Messages API"},{"name":"FlashAttention"},{"name":"FlashInfer"},{"name":"CUTLASS"},{"name":"torch.compile"},{"name":"gRPC"},{"name":"GPTQ"},{"name":"AWQ"},{"name":"GGUF"},{"name":"ModelOpt"},{"name":"TorchAO"},{"name":"OpenAI API"},{"name":"Kubernetes"},{"name":"PyTorch"},{"name":"Ray"},{"name":"OpenTelemetry"},{"name":"Prometheus"},{"name":"FastAPI"},{"name":"Transformers"},{"name":"Outlines"},{"name":"Google Cloud TPU"},{"name":"Intel Gaudi"},{"name":"AMD Instinct"},{"name":"Apple Silicon"},{"name":"IBM Spyre"},{"name":"Huawei Ascend"},{"name":"Rebellions NPU"},{"name":"TRTLLM-GEN"},{"name":"CuTeDSL"},{"name":"OpenAI","native":true,"direction":"outbound"}]` |
| language.primary | Python | Python |
| license.spdx | Apache-2.0 | Apache-2.0 |
| market.availability | - | `{"primaryMarkets":[],"availabilityScope":"global","availableCountries":[],"notAvailableCountries":[]}` |
| platform.support | `{"cli":true,"windows":true}` | `{"cli":true,"linux":true}` |
| pricing | `{"type":"open_source","freeTier":true,"sourceUrl":"https://lmdeploy.readthedocs.io/en/latest/","retrievedAt":"2026-08-20T12:17:08.568Z"}` | `{"type":"open_source","freeTier":true,"sourceUrl":"https://docs.vllm.ai","retrievedAt":"2026-08-14T21:53:56.648Z"}` |
| pricing.free_tier | yes | yes |
| pricing.model | open_source | open_source |
| pricing.price_level | free | free |
| pricing.transparent | - | yes |
| release.cadence_days | 21 | 12 |
| release.history | `[{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.18.0","date":"2026-09-28T07:04:58Z","type":"stable","version":"v0.18.0"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.17.0","date":"2026-09-01T07:13:39Z","type":"stable","version":"v0.17.0"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.16.0","date":"2026-08-19T04:43:53Z","type":"stable","version":"v0.16.0"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.15.0","date":"2026-07-31T13:00:46Z","type":"stable","version":"v0.15.0"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.14.0","date":"2026-06-24T04:36:56Z","type":"stable","version":"v0.14.0"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/0.14.0a2","date":"2026-06-16T04:16:04Z","type":"prerelease","version":"0.14.0a2"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/0.14.0a1","date":"2026-06-01T08:46:02Z","type":"prerelease","version":"0.14.0a1"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.13.0","date":"2026-05-12T03:46:57Z","type":"stable","version":"v0.13.0"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.12.3","date":"2026-04-08T03:37:26Z","type":"stable","version":"v0.12.3"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.12.2","date":"2026-03-18T03:13:55Z","type":"stable","version":"v0.12.2"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.12.1","date":"2026-02-13T09:02:12Z","type":"stable","version":"v0.12.1"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.12.0","date":"2026-02-04T06:28:25Z","type":"stable","version":"v0.12.0"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.11.1","date":"2025-12-24T13:27:06Z","type":"stable","version":"v0.11.1"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.11.0","date":"2025-12-04T06:20:38Z","type":"stable","version":"v0.11.0"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.10.2","date":"2025-10-28T11:32:42Z","type":"stable","version":"v0.10.2"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.10.1","date":"2025-09-26T02:45:30Z","type":"stable","version":"v0.10.1"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.10.0","date":"2025-09-09T05:05:09Z","type":"stable","version":"v0.10.0"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.9.2.post1","date":"2025-08-19T09:44:41Z","type":"stable","version":"v0.9.2.post1"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.9.2","date":"2025-07-26T10:00:25Z","type":"stable","version":"v0.9.2"},{"url":"https://github.com/InternLM/lmdeploy/releases/tag/v0.9.1","date":"2025-07-04T10:05:31Z","type":"stable","version":"v0.9.1"}]` | `[{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.31.0","date":"2026-10-05T06:44:55Z","type":"stable","version":"v0.31.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.30.0","date":"2026-09-22T05:20:54Z","type":"stable","version":"v0.30.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.29.0","date":"2026-09-09T08:54:49Z","type":"stable","version":"v0.29.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.28.0","date":"2026-08-26T09:46:30Z","type":"stable","version":"v0.28.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.27.1","date":"2026-08-11T10:47:49Z","type":"stable","version":"v0.27.1"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.27.0","date":"2026-08-10T21:18:11Z","type":"stable","version":"v0.27.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.26.0","date":"2026-07-27T01:06:58Z","type":"stable","version":"v0.26.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.25.1","date":"2026-07-14T08:51:20Z","type":"stable","version":"v0.25.1"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.25.0","date":"2026-07-11T20:06:44Z","type":"stable","version":"v0.25.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.24.0","date":"2026-06-29T19:41:59Z","type":"stable","version":"v0.24.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.23.0","date":"2026-06-15T05:27:20Z","type":"stable","version":"v0.23.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.22.1","date":"2026-06-05T10:10:00Z","type":"stable","version":"v0.22.1"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.22.0","date":"2026-05-29T10:28:13Z","type":"stable","version":"v0.22.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.21.0","date":"2026-05-15T08:44:26Z","type":"stable","version":"v0.21.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.20.2","date":"2026-05-10T07:37:57Z","type":"stable","version":"v0.20.2"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.20.1","date":"2026-05-04T10:36:26Z","type":"stable","version":"v0.20.1"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.20.0","date":"2026-04-27T21:20:28Z","type":"stable","version":"v0.20.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.19.1","date":"2026-04-18T05:44:42Z","type":"stable","version":"v0.19.1"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.19.0","date":"2026-04-03T02:19:12Z","type":"stable","version":"v0.19.0"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.18.1","date":"2026-03-31T00:53:26Z","type":"stable","version":"v0.18.1"}]` |
| security.trust_center | - | https://docs.vllm.ai/en/latest/usage/security/ |
| security.vulnerabilities | `{"count":10,"source":"https://advisories.ecosyste.ms/api/v1/advisories?ecosystem=pypi&package_name=lmdeploy&per_page=100","last_12m":8,"max_severity":"CRITICAL"}` | `{"count":80,"source":"https://advisories.ecosyste.ms/api/v1/advisories?ecosystem=pypi&package_name=vllm&per_page=100","last_12m":52,"max_severity":"CRITICAL"}` |

## Capabilities (Model Serving)

| Capability | LMDeploy | vLLM |
|---|:--:|:--:|
| **Capabilities** |  |  |
| Continuous batching | ✓ | ✓ |
| Dynamic batching | - | - |
| Multi framework support | - | ✓ |
| GPU acceleration | ✓ | ✓ |
| Multi GPU multi node | ✓ | ✓ |
| Quantization support | ✓ | ✓ |
| Openai compatible API | - | ✓ |
| Autoscaling scale to zero | - | - |
| Multi model serving | ✓ | ✓ |
| Canary ab rollout | - | - |
| Kubernetes native | - | ✓ |
| Open source | - | ✓ |

*Source: vioscaleAI. Generated 2026-10-17T05:43:10.498Z. "-" = undocumented, not absent.*