@@ -21,31 +21,6 @@ export const opencodeGoDefaultModelInfo: ModelInfo = {
2121
2222export const OPENCODE_GO_DEFAULT_TEMPERATURE = 0
2323
24- /**
25- * Limits for Go models that do not yet have a full native configuration below.
26- *
27- * The Go `/v1/models` endpoint currently returns only model IDs, so these
28- * values prevent newly listed models from inheriting the generic 200k context
29- * window. Keep this table aligned with the endpoint and models.dev metadata.
30- */
31- export const opencodeGoModelLimits : Record < string , Pick < ModelInfo , "contextWindow" | "maxTokens" > > = {
32- "kimi-k2.7-code" : { contextWindow : 262_144 , maxTokens : 262_144 } ,
33- "longcat-2.0" : { contextWindow : 1_000_000 , maxTokens : 131_072 } ,
34- "glm-5.3-flash" : { contextWindow : 1_000_000 , maxTokens : 131_072 } ,
35- "deepseek-v4-flash-vision-exp" : { contextWindow : 1_000_000 , maxTokens : 384_000 } ,
36- "qwen3.8-flash" : { contextWindow : 1_000_000 , maxTokens : 131_072 } ,
37- "qwen3.5-plus" : { contextWindow : 262_144 , maxTokens : 65_536 } ,
38- "mimo-v2-pro" : { contextWindow : 1_048_576 , maxTokens : 128_000 } ,
39- "mimo-v2-omni" : { contextWindow : 262_144 , maxTokens : 128_000 } ,
40- "hy4-preview" : { contextWindow : 1_024_000 , maxTokens : 64_000 } ,
41- hy3 : { contextWindow : 256_000 , maxTokens : 64_000 } ,
42- "hy3-preview" : { contextWindow : 256_000 , maxTokens : 64_000 } ,
43- "gpt-5.6-luna" : { contextWindow : 1_050_000 , maxTokens : 128_000 } ,
44- "grok-4.5" : { contextWindow : 500_000 , maxTokens : 500_000 } ,
45- "grok-4.6" : { contextWindow : 500_000 , maxTokens : 500_000 } ,
46- "muse-spark-1.2-contributor" : { contextWindow : 1_048_576 , maxTokens : 131_072 } ,
47- }
48-
4924/**
5025 * Native per-model configuration for the Opencode Go plan.
5126 *
@@ -126,6 +101,21 @@ export const opencodeGoModels: Record<string, ModelInfo> = {
126101 description :
127102 "GLM-5.3 is Zhipu's flagship coding and agent model with a 1M context window, 128k max output, and always-on reasoning with configurable effort (Low/High/Max). Available via the Opencode Go plan." ,
128103 } ,
104+ "glm-5.3-flash" : {
105+ maxTokens : 131_072 ,
106+ contextWindow : 1_000_000 ,
107+ supportsImages : true ,
108+ supportsPromptCache : true ,
109+ supportsMaxTokens : true ,
110+ supportsReasoningEffort : [ "low" , "high" , "max" ] ,
111+ reasoningEffort : "max" ,
112+ preserveReasoning : true ,
113+ inputPrice : 0.075 ,
114+ outputPrice : 0.25 ,
115+ cacheReadsPrice : 0.015 ,
116+ description :
117+ "GLM-5.3 Flash is Zhipu's fast multimodal coding and agent model with a 1M context window and configurable reasoning effort. Available via the Opencode Go plan." ,
118+ } ,
129119 "glm-5.2" : {
130120 maxTokens : 131_072 ,
131121 contextWindow : 1_000_000 ,
@@ -187,6 +177,36 @@ export const opencodeGoModels: Record<string, ModelInfo> = {
187177 description :
188178 "Kimi K2.6 is Moonshot AI's native multimodal agentic MoE model with a 256k context window, built for long-horizon coding and tool use. Available via the Opencode Go plan." ,
189179 } ,
180+ "kimi-k2.7-code" : {
181+ maxTokens : 262_144 ,
182+ contextWindow : 262_144 ,
183+ supportsImages : true ,
184+ supportsPromptCache : true ,
185+ supportsMaxTokens : true ,
186+ preserveReasoning : true ,
187+ defaultTemperature : 1.0 ,
188+ inputPrice : 0.95 ,
189+ outputPrice : 4.0 ,
190+ cacheReadsPrice : 0.19 ,
191+ description :
192+ "Kimi K2.7 Code is Moonshot AI's coding model for long-context programming tasks, with multimodal input and a 256k context window. Available via the Opencode Go plan." ,
193+ } ,
194+
195+ // --- Meituan LongCat ---
196+ "longcat-2.0" : {
197+ maxTokens : 131_072 ,
198+ contextWindow : 1_000_000 ,
199+ supportsImages : false ,
200+ supportsPromptCache : true ,
201+ supportsMaxTokens : true ,
202+ supportsReasoningBinary : true ,
203+ preserveReasoning : true ,
204+ inputPrice : 0.3 ,
205+ outputPrice : 1.2 ,
206+ cacheReadsPrice : 0.006 ,
207+ description :
208+ "LongCat 2.0 is Meituan's long-context reasoning and coding model with a 1M context window. Available via the Opencode Go plan." ,
209+ } ,
190210
191211 // --- Xiaomi MiMo ---
192212 "mimo-v2.5" : {
@@ -225,6 +245,36 @@ export const opencodeGoModels: Record<string, ModelInfo> = {
225245 description :
226246 "MiMo V2.5 Pro - Xiaomi's flagship reasoning model with 1M context, deep thinking, and tool calling. Available via the Opencode Go plan." ,
227247 } ,
248+ "mimo-v2-pro" : {
249+ maxTokens : 128_000 ,
250+ contextWindow : 1_048_576 ,
251+ supportsImages : false ,
252+ supportsPromptCache : false ,
253+ preserveReasoning : true ,
254+ inputPrice : 1.0 ,
255+ outputPrice : 3.0 ,
256+ cacheReadsPrice : 0.2 ,
257+ longContextPricing : {
258+ thresholdTokens : 256_000 ,
259+ inputPriceMultiplier : 2 ,
260+ outputPriceMultiplier : 2 ,
261+ cacheReadsPriceMultiplier : 2 ,
262+ } ,
263+ description :
264+ "MiMo V2 Pro is Xiaomi's text reasoning and coding model with a 1M context window. Available via the Opencode Go plan." ,
265+ } ,
266+ "mimo-v2-omni" : {
267+ maxTokens : 128_000 ,
268+ contextWindow : 262_144 ,
269+ supportsImages : true ,
270+ supportsPromptCache : false ,
271+ preserveReasoning : true ,
272+ inputPrice : 0.4 ,
273+ outputPrice : 2.0 ,
274+ cacheReadsPrice : 0.08 ,
275+ description :
276+ "MiMo V2 Omni is Xiaomi's multimodal reasoning model with a 256k context window. Available via the Opencode Go plan." ,
277+ } ,
228278
229279 // --- MiniMax ---
230280 "minimax-m2.5" : {
@@ -279,6 +329,21 @@ export const opencodeGoModels: Record<string, ModelInfo> = {
279329 } ,
280330
281331 // --- Alibaba Qwen ---
332+ "qwen3.5-plus" : {
333+ maxTokens : 65_536 ,
334+ contextWindow : 262_144 ,
335+ supportsImages : true ,
336+ supportsPromptCache : true ,
337+ supportsReasoningBudget : true ,
338+ supportsReasoningBinary : true ,
339+ preserveReasoning : true ,
340+ inputPrice : 0.2 ,
341+ outputPrice : 1.2 ,
342+ cacheReadsPrice : 0.02 ,
343+ cacheWritesPrice : 0.25 ,
344+ description :
345+ "Qwen3.5 Plus is Alibaba's multimodal reasoning model with a 256k context window. Available via the Opencode Go plan." ,
346+ } ,
282347 "qwen3.6-plus" : {
283348 maxTokens : 65_536 ,
284349 contextWindow : 1_000_000 ,
@@ -345,6 +410,22 @@ export const opencodeGoModels: Record<string, ModelInfo> = {
345410 description :
346411 "Qwen3.8 Max - Alibaba's flagship multimodal reasoning model with a 1M context window, 128k max output, and long-horizon coding and agentic capabilities. Available via the Opencode Go plan." ,
347412 } ,
413+ "qwen3.8-flash" : {
414+ maxTokens : 131_072 ,
415+ contextWindow : 1_000_000 ,
416+ supportsImages : true ,
417+ supportsPromptCache : true ,
418+ supportsMaxTokens : true ,
419+ supportsReasoningBudget : true ,
420+ supportsReasoningBinary : true ,
421+ preserveReasoning : true ,
422+ inputPrice : 0.15 ,
423+ outputPrice : 0.47 ,
424+ cacheReadsPrice : 0.016 ,
425+ cacheWritesPrice : 0.2 ,
426+ description :
427+ "Qwen3.8 Flash is Alibaba's fast multimodal reasoning model with a 1M context window. Available via the Opencode Go plan." ,
428+ } ,
348429
349430 // --- DeepSeek ---
350431 "deepseek-v4-pro" : {
@@ -383,6 +464,147 @@ export const opencodeGoModels: Record<string, ModelInfo> = {
383464 description :
384465 "DeepSeek-V4-Flash is DeepSeek's fast, cost-efficient V4 model supporting thinking and non-thinking modes. Available via the Opencode Go plan." ,
385466 } ,
467+ "deepseek-v4-flash-vision-exp" : {
468+ maxTokens : 384_000 ,
469+ contextWindow : 1_000_000 ,
470+ supportsImages : true ,
471+ supportsPromptCache : true ,
472+ supportsMaxTokens : true ,
473+ supportsReasoningEffort : [ "disable" , "low" , "high" , "max" ] ,
474+ reasoningEffort : "high" ,
475+ preserveReasoning : true ,
476+ inputPrice : 0.22 ,
477+ outputPrice : 0.66 ,
478+ cacheReadsPrice : 0.007 ,
479+ description :
480+ "DeepSeek V4 Flash Vision Experimental is a fast multimodal reasoning model with a 1M context window. Available via the Opencode Go plan." ,
481+ } ,
482+
483+ // --- Tencent Hunyuan ---
484+ "hy4-preview" : {
485+ maxTokens : 64_000 ,
486+ contextWindow : 1_024_000 ,
487+ supportsImages : false ,
488+ supportsPromptCache : true ,
489+ supportsMaxTokens : true ,
490+ supportsReasoningEffort : [ "disable" , "high" ] ,
491+ reasoningEffort : "high" ,
492+ preserveReasoning : true ,
493+ inputPrice : 0.834 ,
494+ outputPrice : 2.501 ,
495+ cacheReadsPrice : 0.042 ,
496+ description :
497+ "Hunyuan 4 Preview is Tencent's long-context reasoning and coding model with a 1M context window. Available via the Opencode Go plan." ,
498+ } ,
499+ hy3 : {
500+ maxTokens : 64_000 ,
501+ contextWindow : 256_000 ,
502+ supportsImages : false ,
503+ supportsPromptCache : true ,
504+ supportsMaxTokens : true ,
505+ supportsReasoningEffort : [ "disable" , "low" , "high" ] ,
506+ reasoningEffort : "high" ,
507+ preserveReasoning : true ,
508+ inputPrice : 0.0175 ,
509+ outputPrice : 0.0725 ,
510+ cacheReadsPrice : 0.004375 ,
511+ description :
512+ "Hunyuan 3 is Tencent's reasoning and coding model with a 256k context window. Available via the Opencode Go plan." ,
513+ } ,
514+ "hy3-preview" : {
515+ // The live endpoint still lists this historical alias; it shares HY3's limits and pricing.
516+ maxTokens : 64_000 ,
517+ contextWindow : 256_000 ,
518+ supportsImages : false ,
519+ supportsPromptCache : true ,
520+ supportsMaxTokens : true ,
521+ supportsReasoningEffort : [ "disable" , "low" , "high" ] ,
522+ reasoningEffort : "high" ,
523+ preserveReasoning : true ,
524+ inputPrice : 0.0175 ,
525+ outputPrice : 0.0725 ,
526+ cacheReadsPrice : 0.004375 ,
527+ description :
528+ "Hunyuan 3 Preview is Tencent's preview reasoning and coding model with a 256k context window. Available via the Opencode Go plan." ,
529+ } ,
530+
531+ // --- Responses API models ---
532+ "gpt-5.6-luna" : {
533+ maxTokens : 128_000 ,
534+ contextWindow : 1_050_000 ,
535+ supportsImages : true ,
536+ supportsPromptCache : true ,
537+ supportsMaxTokens : true ,
538+ supportsReasoningEffort : [ "none" , "low" , "medium" , "high" , "xhigh" , "max" ] ,
539+ reasoningEffort : "medium" ,
540+ inputPrice : 0.2 ,
541+ outputPrice : 1.2 ,
542+ cacheWritesPrice : 0.25 ,
543+ cacheReadsPrice : 0.02 ,
544+ longContextPricing : {
545+ thresholdTokens : 272_000 ,
546+ inputPriceMultiplier : 2 ,
547+ outputPriceMultiplier : 1.5 ,
548+ cacheWritesPriceMultiplier : 2 ,
549+ cacheReadsPriceMultiplier : 2 ,
550+ } ,
551+ description :
552+ "GPT-5.6 Luna is OpenAI's fast reasoning model with a 1M context window. Available via the Opencode Go plan." ,
553+ } ,
554+ "grok-4.5" : {
555+ maxTokens : 500_000 ,
556+ contextWindow : 500_000 ,
557+ supportsImages : true ,
558+ supportsPromptCache : true ,
559+ supportsMaxTokens : true ,
560+ supportsReasoningEffort : [ "low" , "medium" , "high" ] ,
561+ reasoningEffort : "high" ,
562+ inputPrice : 2.0 ,
563+ outputPrice : 6.0 ,
564+ cacheReadsPrice : 0.3 ,
565+ longContextPricing : {
566+ thresholdTokens : 200_000 ,
567+ inputPriceMultiplier : 2 ,
568+ outputPriceMultiplier : 2 ,
569+ cacheReadsPriceMultiplier : 2 ,
570+ } ,
571+ description :
572+ "Grok 4.5 is xAI's multimodal reasoning and agent model with a 500k context window. Available via the Opencode Go plan." ,
573+ } ,
574+ "grok-4.6" : {
575+ maxTokens : 500_000 ,
576+ contextWindow : 500_000 ,
577+ supportsImages : true ,
578+ supportsPromptCache : true ,
579+ supportsMaxTokens : true ,
580+ supportsReasoningEffort : [ "low" , "medium" , "high" , "xhigh" ] ,
581+ reasoningEffort : "high" ,
582+ inputPrice : 2.0 ,
583+ outputPrice : 6.0 ,
584+ cacheReadsPrice : 0.5 ,
585+ longContextPricing : {
586+ thresholdTokens : 200_000 ,
587+ inputPriceMultiplier : 2 ,
588+ outputPriceMultiplier : 2 ,
589+ cacheReadsPriceMultiplier : 2 ,
590+ } ,
591+ description :
592+ "Grok 4.6 is xAI's multimodal reasoning and agent model with a 500k context window. Available via the Opencode Go plan." ,
593+ } ,
594+ "muse-spark-1.2-contributor" : {
595+ maxTokens : 131_072 ,
596+ contextWindow : 1_048_576 ,
597+ supportsImages : true ,
598+ supportsPromptCache : true ,
599+ supportsMaxTokens : true ,
600+ supportsReasoningEffort : [ "minimal" , "low" , "medium" , "high" , "xhigh" ] ,
601+ reasoningEffort : "medium" ,
602+ inputPrice : 0.1 ,
603+ outputPrice : 0.2 ,
604+ cacheReadsPrice : 0.002 ,
605+ description :
606+ "Muse Spark 1.2 Contributor is Meta's multimodal coding model with a 1M context window. Available via the Opencode Go plan." ,
607+ } ,
386608}
387609
388610/**
@@ -404,6 +626,7 @@ export const opencodeGoModels: Record<string, ModelInfo> = {
404626export const OPENCODE_GO_ANTHROPIC_FORMAT_MODELS = new Set < string > ( [
405627 // --- Alibaba Qwen ---
406628 "qwen3.8-max" ,
629+ "qwen3.8-flash" ,
407630 "qwen3.7-max" ,
408631 "qwen3.7-plus" ,
409632 "qwen3.6-plus" ,
@@ -431,11 +654,3 @@ export function isOpencodeGoAnthropicFormatModel(modelId: string): boolean {
431654export function getOpencodeGoModelInfo ( modelId : string ) : ModelInfo | undefined {
432655 return opencodeGoModels [ modelId ]
433656}
434-
435- /** Returns explicit model limits when Go omits them from `/v1/models`. */
436- export function getOpencodeGoModelLimits ( modelId : string ) : Pick < ModelInfo , "contextWindow" | "maxTokens" > | undefined {
437- const native = getOpencodeGoModelInfo ( modelId )
438- return native
439- ? { contextWindow : native . contextWindow , maxTokens : native . maxTokens }
440- : opencodeGoModelLimits [ modelId ]
441- }
0 commit comments