feat(ai): added Kimi K2 models and fixed Claude context window limits

- Added support for Kimi K2, K2 Turbo Preview, and K2.5 models with reasoning capabilities. - Fixed Claude Opus 4.6 context window to 200K across all providers (was incorrectly set to 1M). - Fixed Claude Sonnet 4 context window to 200K across multiple providers (was incorrectly set to 1M). - Improved Kimi Code model fetching to merge fallback models not returned by the API endpoint.
2026-02-10 16:08:53 +01:00
parent 9ef390c09b
commit 0ca5449f3b
3 changed files with 84 additions and 9 deletions
@@ -1,6 +1,14 @@
 # Changelog

 ## [Unreleased]
+### Added
+
+- Added support for Kimi K2, K2 Turbo Preview, and K2.5 models with reasoning capabilities
+
+### Fixed
+
+- Fixed Claude Opus 4.6 context window to 200K across all providers (was incorrectly set to 1M)
+- Fixed Claude Sonnet 4 context window to 200K across multiple providers (was incorrectly set to 1M)

 ## [11.8.0] - 2026-02-10
 ### Added
@@ -249,7 +249,16 @@ async function fetchKimiCodeModels(): Promise<Model<"openai-completions">[]> {
 				});
 			}

-			console.log(`Fetched ${models.length} models from Kimi Code API`);
+			// The /models endpoint only returns "kimi-for-coding" but the API
+			// accepts other model IDs too — merge in fallback models not returned by the API
+			const fetchedIds = new Set(models.map((m) => m.id));
+			const fallbacks = getKimiCodeFallbackModels();
+			for (const fb of fallbacks) {
+				if (!fetchedIds.has(fb.id)) {
+					models.push(fb);
+				}
+			}
+			console.log(`Fetched ${fetchedIds.size} models from Kimi Code API, ${models.length} total with fallbacks`);
 			return models;
 		} catch (error) {
 			console.error("Failed to fetch Kimi Code models:", error);
@@ -1065,7 +1074,8 @@ async function generateModels() {
 			candidate.cost.cacheRead = 0.5;
 			candidate.cost.cacheWrite = 6.25;
 		}
-		if ((candidate.provider === "anthropic" || candidate.provider === "opencode") && candidate.id === "claude-opus-4-6") {
+		// Opus 4.6 1M context is API-only; all providers should use 200K
+		if (candidate.id.includes("opus-4-6") || candidate.id.includes("opus-4.6")) {
 			candidate.contextWindow = 200000;
 		}
 		// opencode lists Claude Sonnet 4/4.5 with 1M context, actual limit is 200K
@@ -121,7 +121,7 @@ export const MODELS = {
 				cacheRead: 0.5,
 				cacheWrite: 6.25,
 			},
-			contextWindow: 1000000,
+			contextWindow: 200000,
 			maxTokens: 128000,
 		} satisfies Model<"bedrock-converse-stream">,
 		"cohere.command-r-plus-v1:0": {
@@ -223,7 +223,7 @@ export const MODELS = {
 				cacheRead: 0.5,
 				cacheWrite: 6.25,
 			},
-			contextWindow: 1000000,
+			contextWindow: 200000,
 			maxTokens: 128000,
 		} satisfies Model<"bedrock-converse-stream">,
 		"eu.anthropic.claude-sonnet-4-20250514-v1:0": {
@@ -325,7 +325,7 @@ export const MODELS = {
 				cacheRead: 0.5,
 				cacheWrite: 6.25,
 			},
-			contextWindow: 1000000,
+			contextWindow: 200000,
 			maxTokens: 128000,
 		} satisfies Model<"bedrock-converse-stream">,
 		"global.anthropic.claude-sonnet-4-20250514-v1:0": {
@@ -920,7 +920,7 @@ export const MODELS = {
 				cacheRead: 0.5,
 				cacheWrite: 6.25,
 			},
-			contextWindow: 1000000,
+			contextWindow: 200000,
 			maxTokens: 128000,
 		} satisfies Model<"bedrock-converse-stream">,
 		"us.anthropic.claude-sonnet-4-20250514-v1:0": {
@@ -1801,7 +1801,7 @@ export const MODELS = {
 				cacheRead: 0,
 				cacheWrite: 0,
 			},
-			contextWindow: 128000,
+			contextWindow: 200000,
 			maxTokens: 64000,
 		} satisfies Model<"openai-completions">,
 		"claude-sonnet-4": {
@@ -3202,6 +3202,63 @@ export const MODELS = {
 			contextWindow: 262144,
 			maxTokens: 32000,
 		} satisfies Model<"openai-completions">,
+		"kimi-k2": {
+			id: "kimi-k2",
+			name: "Kimi K2",
+			api: "openai-completions",
+			provider: "kimi-code",
+			baseUrl: "https://api.kimi.com/coding/v1",
+			headers: {"User-Agent":"KimiCLI/1.0","X-Msh-Platform":"kimi_cli"},
+			compat: {"thinkingFormat":"zai","reasoningContentField":"reasoning_content","supportsDeveloperRole":false},
+			reasoning: true,
+			input: ["text"],
+			cost: {
+				input: 0,
+				output: 0,
+				cacheRead: 0,
+				cacheWrite: 0,
+			},
+			contextWindow: 262144,
+			maxTokens: 32000,
+		} satisfies Model<"openai-completions">,
+		"kimi-k2-turbo-preview": {
+			id: "kimi-k2-turbo-preview",
+			name: "Kimi K2 Turbo Preview",
+			api: "openai-completions",
+			provider: "kimi-code",
+			baseUrl: "https://api.kimi.com/coding/v1",
+			headers: {"User-Agent":"KimiCLI/1.0","X-Msh-Platform":"kimi_cli"},
+			compat: {"thinkingFormat":"zai","reasoningContentField":"reasoning_content","supportsDeveloperRole":false},
+			reasoning: true,
+			input: ["text"],
+			cost: {
+				input: 0,
+				output: 0,
+				cacheRead: 0,
+				cacheWrite: 0,
+			},
+			contextWindow: 262144,
+			maxTokens: 32000,
+		} satisfies Model<"openai-completions">,
+		"kimi-k2.5": {
+			id: "kimi-k2.5",
+			name: "Kimi K2.5",
+			api: "openai-completions",
+			provider: "kimi-code",
+			baseUrl: "https://api.kimi.com/coding/v1",
+			headers: {"User-Agent":"KimiCLI/1.0","X-Msh-Platform":"kimi_cli"},
+			compat: {"thinkingFormat":"zai","reasoningContentField":"reasoning_content","supportsDeveloperRole":false},
+			reasoning: true,
+			input: ["text", "image"],
+			cost: {
+				input: 0,
+				output: 0,
+				cacheRead: 0,
+				cacheWrite: 0,
+			},
+			contextWindow: 262144,
+			maxTokens: 32000,
+		} satisfies Model<"openai-completions">,
 	},
 	"minimax": {
 		"MiniMax-M2": {
@@ -5189,7 +5246,7 @@ export const MODELS = {
 				cacheRead: 0.5,
 				cacheWrite: 6.25,
 			},
-			contextWindow: 1000000,
+			contextWindow: 200000,
 			maxTokens: 128000,
 		} satisfies Model<"openai-completions">,
 		"anthropic/claude-sonnet-4": {
@@ -9189,7 +9246,7 @@ export const MODELS = {
 				cacheRead: 0.5,
 				cacheWrite: 6.25,
 			},
-			contextWindow: 1000000,
+			contextWindow: 200000,
 			maxTokens: 128000,
 		} satisfies Model<"anthropic-messages">,
 		"anthropic/claude-sonnet-4": {