Highlights

• embeddings are now possible to get
• embeddings can be compared with each other
• updated xcframework
• started using claude code

Embeddings

LLM.swift supports text embeddings for semantic similarity and search applications:

:::swift
// Generate embeddings for text
let embeddings1 = try await bot.getEmbeddings("Hello world")
let embeddings2 = try await bot.getEmbeddings("Hi there")
let embeddings3 = try await bot.getEmbeddings("Goodbye")

// Compare similarity (returns 0.0 to 1.0)
let similarity = embeddings1.compare(with: embeddings2)
print(similarity) // 0.8 (high similarity)

// Find most similar embedding
let mostSimilar = embeddings1.findMostSimilar(in: embeddings2, embeddings3)
print(mostSimilar == embeddings2) // true

The Embeddings struct provides: - compare(with:) - Computes cosine similarity between two embeddings (0.0 to 1.0) - findMostSimilar(in:) - Returns the most similar embedding from a set of candidates - Equatable conformance for direct comparison

What's Changed

• Add embeddings support with cosine similarity comparison by @eastriverlee in https://github.com/eastriverlee/LLM.swift/pull/55

Full Changelog: https://github.com/eastriverlee/LLM.swift/compare/v1.7.2...v1.8.0