spring-projects/spring-ai · error · IllegalStateException

EmbeddingModel must be provided

Error message

EmbeddingModel must be provided

What it means

When the builder has no vectorStore, it constructs one from jedisClient + embeddingModel; a null embeddingModel makes construction impossible, so build() throws IllegalStateException 'EmbeddingModel must be provided'. The embedding model is required to vectorize queries for semantic cache lookup.

Source

Thrown at vector-stores/spring-ai-redis-semantic-cache/src/main/java/org/springframework/ai/vectorstore/redis/cache/semantic/DefaultSemanticCache.java:536

		}

		public Builder prefix(String prefix) {
			this.prefix = prefix;
			return this;
		}

		public Builder jedisClient(RedisClient jedisClient) {
			this.jedisClient = jedisClient;
			return this;
		}

		public DefaultSemanticCache build() {
			if (this.vectorStore == null) {
				if (this.jedisClient == null) {
					throw new IllegalStateException("Either vectorStore or jedisClient must be provided");
				}
				if (this.embeddingModel == null) {
					throw new IllegalStateException("EmbeddingModel must be provided");
				}
				this.vectorStore = RedisVectorStore.builder(this.jedisClient, this.embeddingModel)
					.indexName(this.indexName)
					.prefix(this.prefix)
					.metadataFields(MetadataField.text("response"), MetadataField.text("response_text"),
							MetadataField.numeric("ttl"), MetadataField.tag("context_hash"))
					.initializeSchema(true)
					.build();
				if (this.vectorStore instanceof RedisVectorStore redisStore) {
					redisStore.afterPropertiesSet();
				}
			}
			return new DefaultSemanticCache(this.vectorStore, this.similarityThreshold, this.indexName, this.prefix,
					this.useDistanceThreshold);
		}

	}

View on GitHub (pinned to 98a7beda4f)

Solutions

1. Add .embeddingModel(embeddingModel) to the builder chain
2. Ensure an EmbeddingModel bean exists and is injected (e.g. OpenAiEmbeddingModel with an API key)
3. Alternatively supply a fully built vectorStore (which still needs a model elsewhere, but satisfies this check)
4. Fail fast at startup by validating the model bean with an @PostConstruct or configuration check

Example fix

// before
DefaultSemanticCache.builder().jedisClient(jedis).build();
// after
DefaultSemanticCache.builder().jedisClient(jedis).embeddingModel(new OpenAiEmbeddingModel(api)).build();

if (embeddingModel == null) throw new IllegalStateException("EmbeddingModel must be provided to DefaultSemanticCache.builder()");

static boolean hasEmbeddingModel(EmbeddingModel m) { return m != null; }

try { cache = DefaultSemanticCache.builder().jedisClient(jedis).embeddingModel(model).build(); } catch (IllegalStateException e) { /* check bean wiring */ }

When it happens

Trigger: DefaultSemanticCache.builder().jedisClient(jedis).build() without .embeddingModel(...), or passing a null EmbeddingModel bean.

Common situations: Forgetting to wire the EmbeddingModel bean (e.g. OpenAiEmbeddingModel) in Spring config; building the cache in non-Spring code without supplying a model; an optional-injection (@Autowired(required=false)) returning null.

Understand the failure class

Background: "missing required argument" and "the following required arguments were not provided": what required-argument errors mean and how to fix them — this error's family across 20 libraries.

Related errors

• Either vectorStore or jedisClient must be provided
• model cannot be null or empty
• At most one ToolAdvisor is allowed in the advisor chain, but
• Only outputType or outputJsonSchema can be set, not both.
• Multiple tools with the same name (%s) found in ToolCallingC

AI-assisted analysis of spring-projects/spring-ai@98a7beda4f (2026-09-11). Data as JSON: /api/errors/0a42ef64f42329ff. Report an issue: GitHub.