liter_llm 1.1.1 → 1.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 87c0ce7287f3e000b825d496b84662586cb3d8899f11a9f7a8bac55587ddc26c
-  data.tar.gz: 5099baea9360fb98d4949324773b9ef5b3dafc870150f1992f3887162b79994e
+  metadata.gz: 1b6cf155e1ca65bf3dc5166f8f59178a8580cdd9cd6f9f4a1f3dde0ecc5385be
+  data.tar.gz: 07b5f2e4f5ecb3bf1e3bff6dda1cd9c261cb4a6198b727a2ecfe0ac9d97ab55c
 SHA512:
-  metadata.gz: 9452263146c7206ebad3b3ca9b0bc163e251ad832b3b4b4b7352132356e59d29779f2741ebe095607a3cdf7a668ff3eb5cf7ae0b2e6593a043e1f92922761897
-  data.tar.gz: 991f0752533062901b10252157ff0c4fa9e6e69bb69ee5e8218e2124ed70d01cc84c51a1a469ef2a167504b6ff369239712116ed80e1873fb0d662da8c9d3bdd
+  metadata.gz: 8b82f5e04d2eaee3b096c61c199dd9b17463989c1ce11665d65a2258e34fbc92ac545b22d6632d5426275137c33b32c9253382b0025e3a427d82755eea1cf43e
+  data.tar.gz: d530111003fbd0f490264f52e76a2e34ca76072508cd30446b617f58797a2d76b345e029b6ed57301d9c4e475b910ff4cbf3c993f6075ed2cef95b451a84bfbf

data/README.md

@@ -35,6 +35,9 @@
   <a href="https://github.com/kreuzberg-dev/liter-llm/pkgs/container/liter-llm">
     <img src="https://img.shields.io/badge/Docker-007ec6?logo=docker&logoColor=white" alt="Docker">
   </a>
+  <a href="https://github.com/kreuzberg-dev/homebrew-tap/blob/main/Formula/liter-llm.rb">
+    <img src="https://img.shields.io/badge/Homebrew-007ec6?logo=homebrew&logoColor=white" alt="Homebrew">
+  </a>
   <a href="https://github.com/kreuzberg-dev/liter-llm/tree/main/crates/liter-llm-ffi">
     <img src="https://img.shields.io/badge/C-FFI-007ec6" alt="C FFI">
   </a>
@@ -63,7 +66,7 @@
 </div>
 
 
-Universal LLM API client for Ruby. Access 142+ LLM providers through a single interface with idiomatic Ruby API and native performance.
+Universal LLM API client for Ruby. Access 143+ LLM providers through a single interface with idiomatic Ruby API and native performance.
 
 
 ## Installation
@@ -161,7 +164,7 @@ chunks.each { |chunk| puts chunk }
 
 ## Features
 
-### Supported Providers (142+)
+### Supported Providers (143+)
 
 Route to any provider using the `provider/model` prefix convention:
 
@@ -181,7 +184,8 @@ Route to any provider using the `provider/model` prefix convention:
 
 ### Key Capabilities
 
-- **Provider Routing** -- Single client for 142+ LLM providers via `provider/model` prefix
+- **Provider Routing** -- Single client for 143+ LLM providers via `provider/model` prefix
+- **Local LLMs** — Connect to locally-hosted models via Ollama, LM Studio, vLLM, llama.cpp, and other local inference servers
 - **Unified API** -- Consistent `chat`, `chat_stream`, `embeddings`, `list_models` interface
 
 - **Streaming** -- Real-time token streaming via `chat_stream`
@@ -207,7 +211,7 @@ Built on a compiled Rust core for speed and safety:
 
 ## Provider Routing
 
-Route to 142+ providers using the `provider/model` prefix convention:
+Route to 143+ providers using the `provider/model` prefix convention:
 
 ```text
 openai/gpt-4o
@@ -235,7 +239,7 @@ See the [proxy server documentation](https://docs.liter-llm.kreuzberg.dev/server
 
 - **[Documentation](https://docs.liter-llm.kreuzberg.dev)** -- Full docs and API reference
 - **[GitHub Repository](https://github.com/kreuzberg-dev/liter-llm)** -- Source, issues, and discussions
-- **[Provider Registry](https://github.com/kreuzberg-dev/liter-llm/blob/main/schemas/providers.json)** -- 142 supported providers
+- **[Provider Registry](https://github.com/kreuzberg-dev/liter-llm/blob/main/schemas/providers.json)** -- 143 supported providers
 
 Part of [kreuzberg.dev](https://kreuzberg.dev).
 

data/ext/liter_llm_rb/native/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "liter-llm-rb"
-version = "1.1.1"
+version = "1.2.0"
 edition = "2024"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]
 license = "MIT"

data/vendor/Cargo.toml

@@ -2,7 +2,7 @@
 members = ["liter-llm", "liter-llm-ffi"]
 
 [workspace.package]
-version = "1.1.1"
+version = "1.2.0"
 edition = "2024"
 authors = ["Na'aman Hirschfeld <naaman@kreuzberg.dev>"]
 license = "MIT"

data/vendor/liter-llm/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "liter-llm"
-version = "1.1.1"
+version = "1.2.0"
 edition = "2024"
 license = "MIT"
 repository.workspace = true

data/vendor/liter-llm/README.md

@@ -35,6 +35,9 @@
   <a href="https://github.com/kreuzberg-dev/liter-llm/pkgs/container/liter-llm">
     <img src="https://img.shields.io/badge/Docker-007ec6?logo=docker&logoColor=white" alt="Docker">
   </a>
+  <a href="https://github.com/kreuzberg-dev/homebrew-tap/blob/main/Formula/liter-llm.rb">
+    <img src="https://img.shields.io/badge/Homebrew-007ec6?logo=homebrew&logoColor=white" alt="Homebrew">
+  </a>
   <a href="https://github.com/kreuzberg-dev/liter-llm/tree/main/crates/liter-llm-ffi">
     <img src="https://img.shields.io/badge/C-FFI-007ec6" alt="C FFI">
   </a>
@@ -63,7 +66,7 @@
 </div>
 
 
-Universal LLM API client for Rust. Access 142+ LLM providers — OpenAI, Anthropic, Groq, Mistral, and more — through a single unified interface. Async/await with Tokio, streaming via BoxStream, composable Tower middleware stack, and compile-time type safety.
+Universal LLM API client for Rust. Access 143+ LLM providers — OpenAI, Anthropic, Groq, Mistral, and more — through a single unified interface. Async/await with Tokio, streaming via BoxStream, composable Tower middleware stack, and compile-time type safety.
 
 
 ## Installation
@@ -174,7 +177,7 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
 
 ## Features
 
-### Supported Providers (142+)
+### Supported Providers (143+)
 
 Route to any provider using the `provider/model` prefix convention:
 
@@ -194,7 +197,8 @@ Route to any provider using the `provider/model` prefix convention:
 
 ### Key Capabilities
 
-- **Provider Routing** -- Single client for 142+ LLM providers via `provider/model` prefix
+- **Provider Routing** -- Single client for 143+ LLM providers via `provider/model` prefix
+- **Local LLMs** — Connect to locally-hosted models via Ollama, LM Studio, vLLM, llama.cpp, and other local inference servers
 - **Unified API** -- Consistent `chat`, `chat_stream`, `embeddings`, `list_models` interface
 
 - **Streaming** -- Real-time token streaming via `chat_stream`
@@ -220,7 +224,7 @@ Built on a compiled Rust core for speed and safety:
 
 ## Provider Routing
 
-Route to 142+ providers using the `provider/model` prefix convention:
+Route to 143+ providers using the `provider/model` prefix convention:
 
 ```text
 openai/gpt-4o
@@ -248,7 +252,7 @@ See the [proxy server documentation](https://docs.liter-llm.kreuzberg.dev/server
 
 - **[Documentation](https://docs.liter-llm.kreuzberg.dev)** -- Full docs and API reference
 - **[GitHub Repository](https://github.com/kreuzberg-dev/liter-llm)** -- Source, issues, and discussions
-- **[Provider Registry](https://github.com/kreuzberg-dev/liter-llm/blob/main/schemas/providers.json)** -- 142 supported providers
+- **[Provider Registry](https://github.com/kreuzberg-dev/liter-llm/blob/main/schemas/providers.json)** -- 143 supported providers
 
 Part of [kreuzberg.dev](https://kreuzberg.dev).
 

data/vendor/liter-llm/tests/local_llm.rs

@@ -0,0 +1,134 @@
+//! Integration tests against local LLM providers (Ollama).
+//!
+//! These tests require a running Ollama instance with models pulled.
+//! Start with: `task local:up`
+//! Run with: `cargo test -p liter-llm --test local_llm -- --ignored`
+
+use futures_util::StreamExt;
+use liter_llm::{
+    ChatCompletionRequest, ClientConfigBuilder, DefaultClient, EmbeddingInput, EmbeddingRequest, LlmClient,
+};
+
+const OLLAMA_CHAT_MODEL: &str = "ollama/qwen2:0.5b";
+const OLLAMA_EMBED_MODEL: &str = "ollama/all-minilm";
+
+/// Check whether an Ollama instance is reachable.
+async fn is_ollama_available() -> bool {
+    let base = std::env::var("OLLAMA_BASE_URL").unwrap_or_else(|_| "http://localhost:11434".into());
+    reqwest::get(format!("{base}/v1/models")).await.is_ok()
+}
+
+fn ollama_client(model_hint: &str) -> DefaultClient {
+    let config = ClientConfigBuilder::new("").max_retries(2).build();
+    DefaultClient::new(config, Some(model_hint)).expect("failed to build Ollama client")
+}
+
+fn simple_chat_request(model: &str) -> ChatCompletionRequest {
+    serde_json::from_value(serde_json::json!({
+        "model": model,
+        "messages": [{"role": "user", "content": "Say hello in one word."}],
+        "max_tokens": 16,
+    }))
+    .expect("failed to build chat request from JSON")
+}
+
+fn simple_embed_request(model: &str) -> EmbeddingRequest {
+    EmbeddingRequest {
+        model: model.into(),
+        input: EmbeddingInput::Single("hello world".into()),
+        encoding_format: None,
+        dimensions: None,
+        user: None,
+    }
+}
+
+#[tokio::test]
+#[ignore]
+async fn local_chat_ollama() {
+    if !is_ollama_available().await {
+        eprintln!("SKIP: Ollama not available, skipping");
+        return;
+    }
+
+    let client = ollama_client(OLLAMA_CHAT_MODEL);
+    let resp = client.chat(simple_chat_request(OLLAMA_CHAT_MODEL)).await.unwrap();
+
+    assert!(!resp.choices.is_empty(), "should have at least one choice");
+    let choice = &resp.choices[0];
+    assert!(
+        choice.message.content.as_ref().is_some_and(|c| !c.is_empty()),
+        "first choice content should be non-empty"
+    );
+    assert!(choice.finish_reason.is_some(), "finish_reason should be present");
+    assert!(!resp.model.is_empty(), "model field should be non-empty");
+}
+
+#[tokio::test]
+#[ignore]
+async fn local_stream_ollama() {
+    if !is_ollama_available().await {
+        eprintln!("SKIP: Ollama not available, skipping");
+        return;
+    }
+
+    let client = ollama_client(OLLAMA_CHAT_MODEL);
+    let mut stream = client
+        .chat_stream(simple_chat_request(OLLAMA_CHAT_MODEL))
+        .await
+        .unwrap();
+
+    let mut content = String::new();
+    let mut chunk_count = 0u32;
+    let mut saw_finish = false;
+
+    while let Some(result) = stream.next().await {
+        let chunk = result.unwrap();
+        chunk_count += 1;
+        if let Some(choice) = chunk.choices.first() {
+            if let Some(text) = &choice.delta.content {
+                content.push_str(text);
+            }
+            if choice.finish_reason.is_some() {
+                saw_finish = true;
+            }
+        }
+        if chunk_count > 200 {
+            break;
+        }
+    }
+
+    assert!(chunk_count >= 1, "should receive at least 1 chunk");
+    assert!(!content.is_empty(), "concatenated content should be non-empty");
+    assert!(saw_finish, "should see a finish_reason in the stream");
+}
+
+#[tokio::test]
+#[ignore]
+async fn local_embed_ollama() {
+    if !is_ollama_available().await {
+        eprintln!("SKIP: Ollama not available, skipping");
+        return;
+    }
+
+    let client = ollama_client(OLLAMA_EMBED_MODEL);
+    let resp = client.embed(simple_embed_request(OLLAMA_EMBED_MODEL)).await.unwrap();
+
+    assert!(!resp.data.is_empty(), "should have embedding data");
+    assert!(!resp.data[0].embedding.is_empty(), "embedding should have dimensions");
+    assert!(!resp.model.is_empty(), "model field should be non-empty");
+}
+
+#[tokio::test]
+#[ignore]
+async fn local_list_models_ollama() {
+    if !is_ollama_available().await {
+        eprintln!("SKIP: Ollama not available, skipping");
+        return;
+    }
+
+    let client = ollama_client(OLLAMA_CHAT_MODEL);
+    let resp = client.list_models().await.unwrap();
+
+    assert!(!resp.data.is_empty(), "should list at least one model");
+    assert!(!resp.data[0].id.is_empty(), "first model id should be non-empty");
+}

data/vendor/liter-llm-ffi/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "liter-llm-ffi"
-version = "1.1.1"
+version = "1.2.0"
 edition = "2024"
 license = "MIT"
 repository.workspace = true
@@ -20,8 +20,8 @@ default = []
 base64.workspace = true
 bytes.workspace = true
 futures-core.workspace = true
-liter-llm = { path = "../liter-llm", version = "1.1.1", features = ["full"] }
-liter-llm-bindings-core = { path = "../liter-llm-bindings-core", version = "1.1.1" }
+liter-llm = { path = "../liter-llm", version = "1.2.0", features = ["full"] }
+liter-llm-bindings-core = { path = "../liter-llm-bindings-core", version = "1.2.0" }
 serde.workspace = true
 serde_json.workspace = true
 tokio.workspace = true

data/vendor/liter-llm-ffi/liter_llm.h

@@ -8,9 +8,9 @@
 /* Warning, this file is autogenerated by cbindgen. Don't modify this manually. */
 
 #define LITER_LLM_VERSION_MAJOR 1
-#define LITER_LLM_VERSION_MINOR 1
-#define LITER_LLM_VERSION_PATCH 1
-#define LITER_LLM_VERSION "1.1.1"
+#define LITER_LLM_VERSION_MINOR 2
+#define LITER_LLM_VERSION_PATCH 0
+#define LITER_LLM_VERSION "1.2.0"
 
 
 #include <stdarg.h>

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: liter_llm
 version: !ruby/object:Gem::Version
-  version: 1.1.1
+  version: 1.2.0
 platform: ruby
 authors:
 - Na'aman Hirschfeld
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-03-29 00:00:00.000000000 Z
+date: 2026-04-07 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rb_sys
@@ -274,6 +274,7 @@ files:
 - vendor/liter-llm/tests/live_providers/mistral.rs
 - vendor/liter-llm/tests/live_providers/openai.rs
 - vendor/liter-llm/tests/live_providers/vertex_ai.rs
+- vendor/liter-llm/tests/local_llm.rs
 - vendor/liter-llm/tests/middleware_integration.rs
 - vendor/liter-llm/tests/operations_integration.rs
 - vendor/liter-llm/tests/routing_integration.rs