ZeroReader Qwen3 Embedding 0.6B

Choose this endpoint when you need compact, fast embeddings suitable for semantic similarity, clustering, or retrieval-augmented generation tasks, especially in multilingual or mixed-language contexts where Qwen's training data is an advantage. Prefer it over BGE Large EN v1.5 (also on this provider) when you want a smaller, faster model at lower cost, and don't require maximum English-only embedding quality. Good for high-throughput pipelines where embedding speed matters more than peak accuracy.

• Invalid input type (neither string nor array of strings) returns a 400 error
• Empty text input may return a zero vector or error
• Payment failure via x402 protocol results in 402 response blocking the call
• Overly long input texts may be truncated or rejected depending on model token limits
• Network timeout if the model inference takes too long under high load