1package models
 2
 3const (
 4	ProviderGROQ ModelProvider = "groq"
 5
 6	// GROQ
 7	QWENQwq ModelID = "qwen-qwq"
 8
 9	// GROQ preview models
10	Llama4Scout               ModelID = "meta-llama/llama-4-scout-17b-16e-instruct"
11	Llama4Maverick            ModelID = "meta-llama/llama-4-maverick-17b-128e-instruct"
12	Llama3_3_70BVersatile     ModelID = "llama-3.3-70b-versatile"
13	DeepseekR1DistillLlama70b ModelID = "deepseek-r1-distill-llama-70b"
14)
15
16var GroqModels = map[ModelID]Model{
17	//
18	// GROQ
19	QWENQwq: {
20		ID:                 QWENQwq,
21		Name:               "Qwen Qwq",
22		Provider:           ProviderGROQ,
23		APIModel:           "qwen-qwq-32b",
24		CostPer1MIn:        0.29,
25		CostPer1MInCached:  0.275,
26		CostPer1MOutCached: 0.0,
27		CostPer1MOut:       0.39,
28		ContextWindow:      128_000,
29		DefaultMaxTokens:   50000,
30		// for some reason, the groq api doesn't like the reasoningEffort parameter
31		CanReason:           false,
32		SupportsAttachments: false,
33	},
34
35	Llama4Scout: {
36		ID:                  Llama4Scout,
37		Name:                "Llama4Scout",
38		Provider:            ProviderGROQ,
39		APIModel:            "meta-llama/llama-4-scout-17b-16e-instruct",
40		CostPer1MIn:         0.11,
41		CostPer1MInCached:   0,
42		CostPer1MOutCached:  0,
43		CostPer1MOut:        0.34,
44		ContextWindow:       128_000, // 10M when?
45		SupportsAttachments: true,
46	},
47
48	Llama4Maverick: {
49		ID:                  Llama4Maverick,
50		Name:                "Llama4Maverick",
51		Provider:            ProviderGROQ,
52		APIModel:            "meta-llama/llama-4-maverick-17b-128e-instruct",
53		CostPer1MIn:         0.20,
54		CostPer1MInCached:   0,
55		CostPer1MOutCached:  0,
56		CostPer1MOut:        0.20,
57		ContextWindow:       128_000,
58		SupportsAttachments: true,
59	},
60
61	Llama3_3_70BVersatile: {
62		ID:                  Llama3_3_70BVersatile,
63		Name:                "Llama3_3_70BVersatile",
64		Provider:            ProviderGROQ,
65		APIModel:            "llama-3.3-70b-versatile",
66		CostPer1MIn:         0.59,
67		CostPer1MInCached:   0,
68		CostPer1MOutCached:  0,
69		CostPer1MOut:        0.79,
70		ContextWindow:       128_000,
71		SupportsAttachments: false,
72	},
73
74	DeepseekR1DistillLlama70b: {
75		ID:                  DeepseekR1DistillLlama70b,
76		Name:                "DeepseekR1DistillLlama70b",
77		Provider:            ProviderGROQ,
78		APIModel:            "deepseek-r1-distill-llama-70b",
79		CostPer1MIn:         0.75,
80		CostPer1MInCached:   0,
81		CostPer1MOutCached:  0,
82		CostPer1MOut:        0.99,
83		ContextWindow:       128_000,
84		CanReason:           true,
85		SupportsAttachments: false,
86	},
87}