OpenAI o1 model

Cargo.toml

@@ -1,10 +1,11 @@
 [package]
 name = "allms"
-version = "0.13.0"
+version = "0.14.0"
 edition = "2021"
 authors = [
     "Kamil Litman <[email protected]>",
-    "Dario Lencina <[email protected]>",
+    "Dario Lencina <[email protected]>",
+    "Nate Dubetz <[email protected]>"
 ]
 keywords = ["openai", "anthropic", "mistral", "gemini", "assistant"]
 description = "One Library to rule them aLLMs"

README.md

@@ -43,7 +43,9 @@ Mistral:
 
 OpenAI:
 - APIs: Chat Completions, Function Calling, Assistants (v1 & v2), Files, Vector Stores, Tools (file_search)
-- Models: o1 Preview, o1 Mini (Chat Completions only), GPT-4o, GPT-4, GPT-4 32k, GPT-4 Turbo, GPT-3.5 Turbo, GPT-3.5 Turbo 16k, fine-tuned models (via `Custom` variant)
+- Models: 
+    - Chat Completions only: o1, o1 Preview, o1 Mini, o3 Mini 
+    - Chat Completions & Assistants: GPT-4o, GPT-4, GPT-4 32k, GPT-4 Turbo, GPT-3.5 Turbo, GPT-3.5 Turbo 16k, fine-tuned models (via `Custom` variant)
 
 Perplexity:
 - APIs: Chat Completions

examples/use_completions.rs

@@ -44,7 +44,7 @@ async fn main() {
 
     // Get answer using OpenAI
     let openai_api_key: String = std::env::var("OPENAI_API_KEY").expect("OPENAI_API_KEY not set");
-    let model = OpenAIModels::try_from_str("gpt-4o-mini").unwrap_or(OpenAIModels::Gpt4oMini); // Choose the model
+    let model = OpenAIModels::try_from_str("o3-mini").unwrap_or(OpenAIModels::Gpt4oMini); // Choose the model
     println!("OpenAI model: {:#?}", model.as_str());
 
     let openai_completion = Completions::new(model, &openai_api_key, None, None);

src/assistants/openai/openai_assistant.rs

@@ -43,8 +43,14 @@ pub struct OpenAIAssistant {
 }
 
 impl OpenAIAssistant {
-    //Constructor
+    /// Constructor of the Assistant
+    /// Reasoning models are not currently supported in Assistants API. If requested defaulting to `Gpt4o`
     pub fn new(model: OpenAIModels, open_ai_key: &str) -> Self {
+        let model = match model.assistants_support() {
+            false => OpenAIModels::Gpt4o,
+            _ => model,
+        };
+
         OpenAIAssistant {
             id: None,
             thread_id: None,

src/llm_models/openai.rs

@@ -25,8 +25,12 @@ pub enum OpenAIModels {
     Gpt4o,
     Gpt4o20240806,
     Gpt4oMini,
+    // Reasoning models
     O1Preview,
     O1Mini,
+    O1,
+    O3Mini,
+    // Custom models
     Custom { name: String },
 }
 
@@ -47,6 +51,8 @@ impl LLMModel for OpenAIModels {
             OpenAIModels::Gpt4oMini => "gpt-4o-mini",
             OpenAIModels::O1Preview => "o1-preview",
             OpenAIModels::O1Mini => "o1-mini",
+            OpenAIModels::O1 => "o1",
+            OpenAIModels::O3Mini => "o3-mini",
             OpenAIModels::Custom { name } => name.as_str(),
         }
     }
@@ -66,6 +72,8 @@ impl LLMModel for OpenAIModels {
             "gpt-4o-mini" => Some(OpenAIModels::Gpt4oMini),
             "o1-preview" => Some(OpenAIModels::O1Preview),
             "o1-mini" => Some(OpenAIModels::O1Mini),
+            "o1" => Some(OpenAIModels::O1),
+            "o3-mini" => Some(OpenAIModels::O3Mini),
             _ => Some(OpenAIModels::Custom {
                 name: name.to_string(),
             }),
@@ -89,6 +97,8 @@ impl LLMModel for OpenAIModels {
             OpenAIModels::Gpt4oMini => 128_000,
             OpenAIModels::O1Preview => 128_000,
             OpenAIModels::O1Mini => 128_000,
+            OpenAIModels::O1 => 200_000,
+            OpenAIModels::O3Mini => 200_000,
             OpenAIModels::Custom { .. } => 128_000,
         }
     }
@@ -108,6 +118,8 @@ impl LLMModel for OpenAIModels {
             | OpenAIModels::Gpt4_32k
             | OpenAIModels::O1Preview
             | OpenAIModels::O1Mini
+            | OpenAIModels::O1
+            | OpenAIModels::O3Mini
             | OpenAIModels::Custom { .. } => {
                 format!(
                     "{OPENAI_API_URL}/v1/chat/completions",
@@ -136,7 +148,9 @@ impl LLMModel for OpenAIModels {
             | OpenAIModels::Gpt3_5Turbo
             | OpenAIModels::Gpt4_32k
             | OpenAIModels::O1Preview
-            | OpenAIModels::O1Mini => false,
+            | OpenAIModels::O1
+            | OpenAIModels::O1Mini
+            | OpenAIModels::O3Mini => false,
             OpenAIModels::Gpt3_5Turbo0613
             | OpenAIModels::Gpt3_5Turbo16k
             | OpenAIModels::Gpt4
@@ -256,7 +270,10 @@ impl LLMModel for OpenAIModels {
             // - Tools: tools, function calling, and response format parameters are not supported.
             // - Other: temperature, top_p and n are fixed at 1, while presence_penalty and frequency_penalty are fixed at 0.
             // - Assistants and Batch: these models are not supported in the Assistants API or Batch API.
-            OpenAIModels::O1Preview | OpenAIModels::O1Mini => {
+            OpenAIModels::O1Preview
+            | OpenAIModels::O1Mini
+            | OpenAIModels::O1
+            | OpenAIModels::O3Mini => {
                 let base_instructions = self.get_base_instructions(Some(function_call));
                 let system_message = json!({
                     "role": "user",
@@ -352,6 +369,8 @@ impl LLMModel for OpenAIModels {
             | OpenAIModels::Gpt4_32k
             | OpenAIModels::O1Preview
             | OpenAIModels::O1Mini
+            | OpenAIModels::O1
+            | OpenAIModels::O3Mini
             | OpenAIModels::Custom { .. } => {
                 //Convert API response to struct representing expected response format
                 let chat_response: OpenAPIChatResponse = serde_json::from_str(response_text)?;
@@ -425,7 +444,6 @@ impl LLMModel for OpenAIModels {
                 tpm: 1_000_000,
                 rpm: 10_000,
             },
-            // https://help.openai.com/en/articles/9824962-openai-o1-preview-and-o1-mini-usage-limits-on-chatgpt-and-the-api
             OpenAIModels::O1Preview => RateLimit {
                 tpm: 30_000_000,
                 rpm: 10_000,
@@ -434,6 +452,14 @@ impl LLMModel for OpenAIModels {
                 tpm: 150_000_000,
                 rpm: 30_000,
             },
+            OpenAIModels::O1 => RateLimit {
+                tpm: 30_000_000,
+                rpm: 1_000,
+            },
+            OpenAIModels::O3Mini => RateLimit {
+                tpm: 150_000_000,
+                rpm: 30_000,
+            },
             OpenAIModels::TextDavinci003 => RateLimit {
                 tpm: 250_000,
                 rpm: 3_000,
@@ -476,6 +502,18 @@ impl OpenAIModels {
                 | OpenAIModels::Custom { .. }
         )
     }
+
+    // This function checks if a model supports use in Assistants API
+    // Reasoning models are NOT currently supported
+    pub fn assistants_support(&self) -> bool {
+        !matches!(
+            self,
+            OpenAIModels::O1Preview
+                | OpenAIModels::O1Mini
+                | OpenAIModels::O1
+                | OpenAIModels::O3Mini
+        )
+    }
 }
 
 #[cfg(test)]