xinyin025/one-api
1
0
Fork 0
mirror of https://github.com/songquanpeng/one-api.git synced 2025-09-17 09:16:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

feat: support gemini-2.0-flash

Browse Source 
- Enhance model support by adding new entries and refining checks for system instruction compatibility.
- Update logging display behavior and adjust default quotas for better user experience.
- Revamp pricing structures in the billing system to reflect current model values and deprecate outdated entries.
- Streamline code by replacing hardcoded values with configurations for maintainability.
This commit is contained in:
Laisky.Cai 2025-02-06 03:08:41 +00:00
parent d2903b673d
commit d70335ecea
7 changed files with 71 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
relay/adaptor/gemini/adaptor.go
View File

@ -5,8 +5,10 @@ import (
"fmt"
"io"
"net/http"
"strings"
"github.com/gin-gonic/gin"
"github.com/songquanpeng/one-api/common/config"
"github.com/songquanpeng/one-api/common/helper"
channelhelper "github.com/songquanpeng/one-api/relay/adaptor"
"github.com/songquanpeng/one-api/relay/adaptor/openai"
@ -19,15 +21,11 @@ type Adaptor struct {
}
func (a *Adaptor) Init(meta *meta.Meta) {
}
func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
var defaultVersion string
switch meta.ActualModelName {
case "gemini-2.0-flash-exp",
"gemini-2.0-flash-thinking-exp",
"gemini-2.0-flash-thinking-exp-01-21":
defaultVersion := config.GeminiVersion
if strings.Contains(meta.ActualModelName, "-2.0") {
defaultVersion = "v1beta"
}

36
relay/adaptor/gemini/constants.go
View File

@ -4,8 +4,38 @@ 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-flash-thinking-exp", "gemini-2.0-flash-thinking-exp-01-21",
"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",
"gemini-2.0-pro-exp-02-05",
}
// ModelsSupportSystemInstruction is the list of models that support system instruction.
//
// https://cloud.google.com/vertex-ai/generative-ai/docs/learn/prompts/system-instructions
var ModelsSupportSystemInstruction = []string{
// "gemini-1.0-pro-002",
// "gemini-1.5-flash", "gemini-1.5-flash-001", "gemini-1.5-flash-002",
// "gemini-1.5-flash-8b",
// "gemini-1.5-pro", "gemini-1.5-pro-001", "gemini-1.5-pro-002",
// "gemini-1.5-pro-experimental",
"gemini-2.0-flash", "gemini-2.0-flash-exp",
"gemini-2.0-flash-thinking-exp-01-21",
}
// IsModelSupportSystemInstruction check if the model support system instruction.
//
// Because the main version of Go is 1.20, slice.Contains cannot be used
func IsModelSupportSystemInstruction(model string) bool {
for _, m := range ModelsSupportSystemInstruction {
if m == model {
return true
}
}
return false
}

9
relay/adaptor/gemini/main.go
View File

@ -132,9 +132,16 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *ChatRequest {
}
// Converting system prompt to prompt from user for the same reason
if content.Role == "system" {
content.Role = "user"
shouldAddDummyModelMessage = true
if IsModelSupportSystemInstruction(textRequest.Model) {
geminiRequest.SystemInstruction = &content
geminiRequest.SystemInstruction.Role = ""
continue
} else {
content.Role = "user"
}
}
geminiRequest.Contents = append(geminiRequest.Contents, content)
// If a system message is the last message, we need to add a dummy model message to make gemini happy

9
relay/adaptor/gemini/model.go
View File

@ -1,10 +1,11 @@
package gemini
type ChatRequest struct {
Contents []ChatContent `json:"contents"`
SafetySettings []ChatSafetySettings `json:"safety_settings,omitempty"`
GenerationConfig ChatGenerationConfig `json:"generation_config,omitempty"`
Tools []ChatTools `json:"tools,omitempty"`
Contents []ChatContent `json:"contents"`
SafetySettings []ChatSafetySettings `json:"safety_settings,omitempty"`
GenerationConfig ChatGenerationConfig `json:"generation_config,omitempty"`
Tools []ChatTools `json:"tools,omitempty"`
SystemInstruction *ChatContent `json:"system_instruction,omitempty"`
}
type EmbeddingRequest struct {

30
relay/billing/ratio/model.go
View File

@ -88,11 +88,11 @@ var ModelRatio = map[string]float64{
"claude-2.1": 8.0 / 1000 * USD,
"claude-3-haiku-20240307": 0.25 / 1000 * USD,
"claude-3-5-haiku-20241022": 1.0 / 1000 * USD,
"claude-3-5-haiku-latest": 1.0 / 1000 * USD,
"claude-3-5-haiku-latest": 1.0 / 1000 * USD,
"claude-3-sonnet-20240229": 3.0 / 1000 * USD,
"claude-3-5-sonnet-20240620": 3.0 / 1000 * USD,
"claude-3-5-sonnet-20241022": 3.0 / 1000 * USD,
"claude-3-5-sonnet-latest" : 3.0 / 1000 * USD,
"claude-3-5-sonnet-latest": 3.0 / 1000 * USD,
"claude-3-opus-20240229": 15.0 / 1000 * USD,
// https://cloud.baidu.com/doc/WENXINWORKSHOP/s/hlrk4akp7
"ERNIE-4.0-8K": 0.120 * RMB,
@ -112,15 +112,23 @@ var ModelRatio = map[string]float64{
"bge-large-en": 0.002 * RMB,
"tao-8k": 0.002 * RMB,
// https://ai.google.dev/pricing
"gemini-pro": 1, // $0.00025 / 1k characters -> $0.001 / 1k tokens
"gemini-1.0-pro": 1,
"gemini-1.5-pro": 1,
"gemini-1.5-pro-001": 1,
"gemini-1.5-flash": 1,
"gemini-1.5-flash-001": 1,
"gemini-2.0-flash-exp": 1,
"gemini-2.0-flash-thinking-exp": 1,
"gemini-2.0-flash-thinking-exp-01-21": 1,
// 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,

2
web/default/src/components/LogsTable.js
View File

@ -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>
</>
)}

2
web/default/src/helpers/utils.js
View File

@ -224,4 +224,4 @@ export function getChannelModels(type) {
return channelModels[type];
}
return [];
}
}