feat: support gemini-2.0-flash

relay/adaptor/gemini/adaptor.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"io"
 	"net/http"
+	"strings"
 
 	"github.com/gin-gonic/gin"
 	"github.com/pkg/errors"
@@ -24,13 +25,9 @@ func (a *Adaptor) Init(meta *meta.Meta) {
 }
 
 func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
-	var defaultVersion string
+	defaultVersion := config.GeminiVersion
-	switch meta.ActualModelName {
+	if strings.Contains(meta.ActualModelName, "-2.0") {
-	case "gemini-2.0-flash-exp",
-		"gemini-2.0-flash-thinking-exp-01-21":
 		defaultVersion = "v1beta"
-	default:
-		defaultVersion = config.GeminiVersion
 	}
 
 	version := helper.AssignOrDefault(meta.Config.APIVersion, defaultVersion)

relay/adaptor/gemini/constants.go

@@ -4,8 +4,12 @@ package gemini
 
 var ModelList = []string{
 	"gemini-pro", "gemini-1.0-pro",
-	"gemini-1.5-flash", "gemini-1.5-pro",
+	// "gemma-2-2b-it", "gemma-2-9b-it", "gemma-2-27b-it",
+	"gemini-1.5-flash", "gemini-1.5-flash-8b",
+	"gemini-1.5-pro", "gemini-1.5-pro-experimental",
 	"text-embedding-004", "aqa",
-	"gemini-2.0-flash-exp",
+	"gemini-2.0-pro-exp-02-05",
+	"gemini-2.0-flash", "gemini-2.0-flash-exp",
+	"gemini-2.0-flash-lite-preview-02-05",
 	"gemini-2.0-flash-thinking-exp-01-21",
 }

relay/adaptor/gemini/main.go

@@ -160,6 +160,15 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *ChatRequest {
 		geminiRequest.Contents = append(geminiRequest.Contents, content)
 	}
 
+	// As of 2025-02-06, the newly released gemini 2.0 models do not support system_instruction,
+	// which can reasonably be considered a bug. Google may fix this issue in the future.
+	if geminiRequest.SystemInstruction != nil &&
+		strings.Contains(textRequest.Model, "-2.0") &&
+		textRequest.Model != "gemini-2.0-flash-exp" &&
+		textRequest.Model != "gemini-2.0-flash-thinking-exp-01-21" {
+		geminiRequest.SystemInstruction = nil
+	}
+
 	return &geminiRequest
 }
 

relay/billing/ratio/model.go

@@ -122,14 +122,23 @@ var ModelRatio = map[string]float64{
 	"bge-large-en":       0.002 * RMB,
 	"tao-8k":             0.002 * RMB,
 	// https://ai.google.dev/pricing
+	// https://cloud.google.com/vertex-ai/generative-ai/pricing
+	// "gemma-2-2b-it":                       0,
+	// "gemma-2-9b-it":                       0,
+	// "gemma-2-27b-it":                      0,
 	"gemini-pro":                          0.25 * MILLI_USD, // $0.00025 / 1k characters -> $0.001 / 1k tokens
 	"gemini-1.0-pro":                      0.125 * MILLI_USD,
 	"gemini-1.5-pro":                      1.25 * MILLI_USD,
 	"gemini-1.5-pro-001":                  1.25 * MILLI_USD,
+	"gemini-1.5-pro-experimental":         1.25 * MILLI_USD,
 	"gemini-1.5-flash":                    0.075 * MILLI_USD,
 	"gemini-1.5-flash-001":                0.075 * MILLI_USD,
+	"gemini-1.5-flash-8b":                 0.0375 * MILLI_USD,
 	"gemini-2.0-flash-exp":                0.075 * MILLI_USD,
+	"gemini-2.0-flash":                    0.15 * MILLI_USD,
+	"gemini-2.0-flash-lite-preview-02-05": 0.075 * MILLI_USD,
 	"gemini-2.0-flash-thinking-exp-01-21": 0.075 * MILLI_USD,
+	"gemini-2.0-pro-exp-02-05":            1.25 * MILLI_USD,
 	"aqa":                                 1,
 	// https://open.bigmodel.cn/pricing
 	"glm-zero-preview": 0.01 * RMB,

web/default/src/components/LogsTable.js

@@ -58,7 +58,7 @@ function renderType(type) {
     case 2:
       return (
         <Label basic color='olive'>
-          Consumption
+          Consumed
         </Label>
       );
     case 3:
@@ -564,7 +564,7 @@ const LogsTable = () => {
                         {log.completion_tokens ? log.completion_tokens : ''}
                       </Table.Cell>
                       <Table.Cell>
-                        {log.quota ? renderQuota(log.quota, t, 6) : ''}
+                        {log.quota ? renderQuota(log.quota, t, 6) : 'free'}
                       </Table.Cell>
                     </>
                   )}