feat: Updated OpenAPI spec

src/libs/Mistral/Generated/Mistral.ChatClient.ChatCompletion.g.cs

@@ -100,6 +100,7 @@ partial void ProcessChatCompletionResponseContent(
                 PromptMode = request.PromptMode,
                 ReasoningEffort = request.ReasoningEffort,
                 Guardrails = request.Guardrails,
+                PromptCacheKey = request.PromptCacheKey,
                 SafePrompt = request.SafePrompt,
             };
             PrepareArguments(
@@ -562,6 +563,9 @@ partial void ProcessChatCompletionResponseContent(
         /// A list of guardrail configurations to apply to this request. Each guardrail specifies a moderation type, categories with thresholds to evaluate, and an action to take on violation.<br/>
         /// Default Value: openapi-json-null-sentinel-value-2BF93600-0FE4-4250-987A-E5DDB203E464
         /// </param>
+        /// <param name="promptCacheKey">
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </param>
         /// <param name="safePrompt">
         /// Whether to inject a safety prompt before all conversations.<br/>
         /// Default Value: false
@@ -589,6 +593,7 @@ partial void ProcessChatCompletionResponseContent(
             global::Mistral.MistralPromptMode? promptMode = default,
             global::Mistral.ChatCompletionRequestReasoningEffort? reasoningEffort = default,
             global::System.Collections.Generic.IList<global::Mistral.GuardrailConfig>? guardrails = default,
+            string? promptCacheKey = default,
             bool? safePrompt = default,
             global::Mistral.AutoSDKRequestOptions? requestOptions = default,
             global::System.Threading.CancellationToken cancellationToken = default)
@@ -615,6 +620,7 @@ partial void ProcessChatCompletionResponseContent(
                 PromptMode = promptMode,
                 ReasoningEffort = reasoningEffort,
                 Guardrails = guardrails,
+                PromptCacheKey = promptCacheKey,
                 SafePrompt = safePrompt,
             };
 

src/libs/Mistral/Generated/Mistral.ChatClient.ChatCompletionAsStream.g.cs

@@ -73,6 +73,7 @@ partial void ProcessChatCompletionAsStreamResponse(
                 PromptMode = request.PromptMode,
                 ReasoningEffort = request.ReasoningEffort,
                 Guardrails = request.Guardrails,
+                PromptCacheKey = request.PromptCacheKey,
                 SafePrompt = request.SafePrompt,
             };
             PrepareArguments(
@@ -467,6 +468,9 @@ partial void ProcessChatCompletionAsStreamResponse(
         /// A list of guardrail configurations to apply to this request. Each guardrail specifies a moderation type, categories with thresholds to evaluate, and an action to take on violation.<br/>
         /// Default Value: openapi-json-null-sentinel-value-2BF93600-0FE4-4250-987A-E5DDB203E464
         /// </param>
+        /// <param name="promptCacheKey">
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </param>
         /// <param name="safePrompt">
         /// Whether to inject a safety prompt before all conversations.<br/>
         /// Default Value: false
@@ -494,6 +498,7 @@ partial void ProcessChatCompletionAsStreamResponse(
             global::Mistral.MistralPromptMode? promptMode = default,
             global::Mistral.ChatCompletionRequestReasoningEffort? reasoningEffort = default,
             global::System.Collections.Generic.IList<global::Mistral.GuardrailConfig>? guardrails = default,
+            string? promptCacheKey = default,
             bool? safePrompt = default,
             global::Mistral.AutoSDKRequestOptions? requestOptions = default,
             [global::System.Runtime.CompilerServices.EnumeratorCancellation] global::System.Threading.CancellationToken cancellationToken = default)
@@ -520,6 +525,7 @@ partial void ProcessChatCompletionAsStreamResponse(
                 PromptMode = promptMode,
                 ReasoningEffort = reasoningEffort,
                 Guardrails = guardrails,
+                PromptCacheKey = promptCacheKey,
                 SafePrompt = safePrompt,
             };
 

src/libs/Mistral/Generated/Mistral.FimClient.FimCompletion.g.cs

@@ -93,6 +93,7 @@ partial void ProcessFimCompletionResponseContent(
                 Prompt = request.Prompt,
                 Suffix = request.Suffix,
                 MinTokens = request.MinTokens,
+                PromptCacheKey = request.PromptCacheKey,
             };
             PrepareArguments(
                 client: HttpClient);
@@ -523,6 +524,9 @@ partial void ProcessFimCompletionResponseContent(
         /// <param name="minTokens">
         /// The minimum number of tokens to generate in the completion.
         /// </param>
+        /// <param name="promptCacheKey">
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </param>
         /// <param name="requestOptions">Per-request overrides such as headers, query parameters, timeout, retries, and response buffering.</param>
         /// <param name="cancellationToken">The token to cancel the operation with</param>
         /// <exception cref="global::System.InvalidOperationException"></exception>
@@ -537,6 +541,7 @@ partial void ProcessFimCompletionResponseContent(
             object? metadata = default,
             string? suffix = default,
             int? minTokens = default,
+            string? promptCacheKey = default,
             global::Mistral.AutoSDKRequestOptions? requestOptions = default,
             global::System.Threading.CancellationToken cancellationToken = default)
         {
@@ -553,6 +558,7 @@ partial void ProcessFimCompletionResponseContent(
                 Prompt = prompt,
                 Suffix = suffix,
                 MinTokens = minTokens,
+                PromptCacheKey = promptCacheKey,
             };
 
             return await FimCompletionAsync(

src/libs/Mistral/Generated/Mistral.FimClient.FimCompletionAsStream.g.cs

@@ -65,6 +65,7 @@ partial void ProcessFimCompletionAsStreamResponse(
                 Prompt = request.Prompt,
                 Suffix = request.Suffix,
                 MinTokens = request.MinTokens,
+                PromptCacheKey = request.PromptCacheKey,
             };
             PrepareArguments(
                 client: HttpClient);
@@ -427,6 +428,9 @@ partial void ProcessFimCompletionAsStreamResponse(
         /// <param name="minTokens">
         /// The minimum number of tokens to generate in the completion.
         /// </param>
+        /// <param name="promptCacheKey">
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </param>
         /// <param name="requestOptions">Per-request overrides such as headers, query parameters, timeout, retries, and response buffering.</param>
         /// <param name="cancellationToken">The token to cancel the operation with</param>
         /// <exception cref="global::System.InvalidOperationException"></exception>
@@ -441,6 +445,7 @@ partial void ProcessFimCompletionAsStreamResponse(
             object? metadata = default,
             string? suffix = default,
             int? minTokens = default,
+            string? promptCacheKey = default,
             global::Mistral.AutoSDKRequestOptions? requestOptions = default,
             [global::System.Runtime.CompilerServices.EnumeratorCancellation] global::System.Threading.CancellationToken cancellationToken = default)
         {
@@ -457,6 +462,7 @@ partial void ProcessFimCompletionAsStreamResponse(
                 Prompt = prompt,
                 Suffix = suffix,
                 MinTokens = minTokens,
+                PromptCacheKey = promptCacheKey,
             };
 
             var __enumerable = FimCompletionAsStreamAsync(

src/libs/Mistral/Generated/Mistral.IChatClient.ChatCompletion.g.cs

@@ -93,6 +93,9 @@ public partial interface IChatClient
         /// A list of guardrail configurations to apply to this request. Each guardrail specifies a moderation type, categories with thresholds to evaluate, and an action to take on violation.<br/>
         /// Default Value: openapi-json-null-sentinel-value-2BF93600-0FE4-4250-987A-E5DDB203E464
         /// </param>
+        /// <param name="promptCacheKey">
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </param>
         /// <param name="safePrompt">
         /// Whether to inject a safety prompt before all conversations.<br/>
         /// Default Value: false
@@ -120,6 +123,7 @@ public partial interface IChatClient
             global::Mistral.MistralPromptMode? promptMode = default,
             global::Mistral.ChatCompletionRequestReasoningEffort? reasoningEffort = default,
             global::System.Collections.Generic.IList<global::Mistral.GuardrailConfig>? guardrails = default,
+            string? promptCacheKey = default,
             bool? safePrompt = default,
             global::Mistral.AutoSDKRequestOptions? requestOptions = default,
             global::System.Threading.CancellationToken cancellationToken = default);

src/libs/Mistral/Generated/Mistral.IChatClient.ChatCompletionAsStream.g.cs

@@ -81,6 +81,9 @@ public partial interface IChatClient
         /// A list of guardrail configurations to apply to this request. Each guardrail specifies a moderation type, categories with thresholds to evaluate, and an action to take on violation.<br/>
         /// Default Value: openapi-json-null-sentinel-value-2BF93600-0FE4-4250-987A-E5DDB203E464
         /// </param>
+        /// <param name="promptCacheKey">
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </param>
         /// <param name="safePrompt">
         /// Whether to inject a safety prompt before all conversations.<br/>
         /// Default Value: false
@@ -108,6 +111,7 @@ public partial interface IChatClient
             global::Mistral.MistralPromptMode? promptMode = default,
             global::Mistral.ChatCompletionRequestReasoningEffort? reasoningEffort = default,
             global::System.Collections.Generic.IList<global::Mistral.GuardrailConfig>? guardrails = default,
+            string? promptCacheKey = default,
             bool? safePrompt = default,
             global::Mistral.AutoSDKRequestOptions? requestOptions = default,
             global::System.Threading.CancellationToken cancellationToken = default);

src/libs/Mistral/Generated/Mistral.IFimClient.FimCompletion.g.cs

@@ -64,6 +64,9 @@ public partial interface IFimClient
         /// <param name="minTokens">
         /// The minimum number of tokens to generate in the completion.
         /// </param>
+        /// <param name="promptCacheKey">
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </param>
         /// <param name="requestOptions">Per-request overrides such as headers, query parameters, timeout, retries, and response buffering.</param>
         /// <param name="cancellationToken">The token to cancel the operation with</param>
         /// <exception cref="global::System.InvalidOperationException"></exception>
@@ -78,6 +81,7 @@ public partial interface IFimClient
             object? metadata = default,
             string? suffix = default,
             int? minTokens = default,
+            string? promptCacheKey = default,
             global::Mistral.AutoSDKRequestOptions? requestOptions = default,
             global::System.Threading.CancellationToken cancellationToken = default);
     }

src/libs/Mistral/Generated/Mistral.IFimClient.FimCompletionAsStream.g.cs

@@ -51,6 +51,9 @@ public partial interface IFimClient
         /// <param name="minTokens">
         /// The minimum number of tokens to generate in the completion.
         /// </param>
+        /// <param name="promptCacheKey">
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </param>
         /// <param name="requestOptions">Per-request overrides such as headers, query parameters, timeout, retries, and response buffering.</param>
         /// <param name="cancellationToken">The token to cancel the operation with</param>
         /// <exception cref="global::System.InvalidOperationException"></exception>
@@ -65,6 +68,7 @@ public partial interface IFimClient
             object? metadata = default,
             string? suffix = default,
             int? minTokens = default,
+            string? promptCacheKey = default,
             global::Mistral.AutoSDKRequestOptions? requestOptions = default,
             global::System.Threading.CancellationToken cancellationToken = default);
     }

src/libs/Mistral/Generated/Mistral.Models.AgentsCompletionRequest.g.cs

@@ -112,6 +112,12 @@ public sealed partial class AgentsCompletionRequest
         [global::System.Text.Json.Serialization.JsonConverter(typeof(global::Mistral.JsonConverters.AgentsCompletionRequestReasoningEffortJsonConverter))]
         public global::Mistral.AgentsCompletionRequestReasoningEffort? ReasoningEffort { get; set; }
 
+        /// <summary>
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </summary>
+        [global::System.Text.Json.Serialization.JsonPropertyName("prompt_cache_key")]
+        public string? PromptCacheKey { get; set; }
+
         /// <summary>
         /// The ID of the agent to use for this completion.
         /// </summary>
@@ -179,6 +185,9 @@ public sealed partial class AgentsCompletionRequest
         /// <param name="reasoningEffort">
         /// Controls the reasoning effort level for reasoning models. "high" enables comprehensive reasoning traces, "none" disables reasoning effort.
         /// </param>
+        /// <param name="promptCacheKey">
+        /// A cache key to enable prompt caching. When provided, the API will attempt to reuse previously computed tokens for requests sharing the same prefix (e.g. multi-turn conversations or requests with a similar system prompt). Cached tokens are billed at 10% of the standard input token price.
+        /// </param>
 #if NET7_0_OR_GREATER
         [global::System.Diagnostics.CodeAnalysis.SetsRequiredMembers]
 #endif
@@ -199,7 +208,8 @@ public AgentsCompletionRequest(
             global::Mistral.Prediction? prediction,
             bool? parallelToolCalls,
             global::Mistral.MistralPromptMode? promptMode,
-            global::Mistral.AgentsCompletionRequestReasoningEffort? reasoningEffort)
+            global::Mistral.AgentsCompletionRequestReasoningEffort? reasoningEffort,
+            string? promptCacheKey)
         {
             this.MaxTokens = maxTokens;
             this.Stream = stream;
@@ -217,6 +227,7 @@ public AgentsCompletionRequest(
             this.ParallelToolCalls = parallelToolCalls;
             this.PromptMode = promptMode;
             this.ReasoningEffort = reasoningEffort;
+            this.PromptCacheKey = promptCacheKey;
             this.AgentId = agentId ?? throw new global::System.ArgumentNullException(nameof(agentId));
         }