vllm-embedding Helm chart
kubeblocksVerified publisherNot scored yet
Self-hosted multimodal embedding service on vLLM for ApeMind private deployment. Serves an OpenAI-compatible /v1/embeddings endpoint (vLLM --runner pooling). Default model pin: Qwen/Qwen3-VL-Embedding-2B (2026-01, Apache-2.0, SOTA multimodal embeddings, MRL dims 64..2048, ~6G VRAM — fits beside other services on a 24G card). Chat/agent LLM is deploy/vllm-chat; the MinerU document-parsing VLM is deploy/mineru openaiServer.
Latest 0.1.0 1 month agoapp version 0.21.0 (not verified against the render) 0Artifact Hub
Radar Score
Latest render attempt (7 Sept 2026) failed: Error: execution error at (vllm-embedding/templates/deployment.yaml:47:20):
.Values.image.repository is REQUIRED. Use vllm/vllm-openai (connected clusters) or your private air-gap mirror of it. See values.yaml image section.
Use --debug flag to render out invalid YAML
Not yet scored
The daily scoring run has not folded the 0 images into a score yet.
Radar Score is the sum of every finding’s contribution; the colour is the worst finding’s band.
Container images
The render produced no pod templates.
Rendered with the chart’s default values. Optional subcharts, images set by operator flags and images inside CRD payloads are not seen and are counted as unmeasured.
Vulnerabilities
No current finding on any measured image of this version.
Indexed versions
| Version | Published | App version | Vulnerabilities | Radar Score |
|---|---|---|---|---|
| 0.1.0latest | 1 month ago | 0.21.0 | — | — |
The latest version and the previous major, as selected nightly from the repository’s index.
README badge
Markdown. The badge shows the latest version’s band and Radar Score, refreshed daily.