Add Vertex AI express mode support (#3977)

* Add Vertex AI express mode support
Split Google AI Studio and Vertex AI

* Add support for Vertex AI, including updating default models and related settings, modifying frontend HTML to include Vertex AI options, and adjusting request processing logic in the backend API.

* Log API name in the console

* Merge sysprompt toggles back

* Use Gemma tokenizers for Vertex and LearnLM

* AI Studio parity updates

* Add link to express mode doc. Also technically it's not a form

* Split title

* Use array includes

* Add support for Google Vertex AI in image captioning feature

* Specify caption API name, add to compression list

---------

Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
This commit is contained in:
NijikaMyWaifu
2025-05-23 01:10:53 +08:00
committed by GitHub
parent 6dc59b9fd3
commit 157315cd68
18 changed files with 193 additions and 42 deletions

View File

@@ -414,6 +414,7 @@ jQuery(async function () {
'openai': SECRET_KEYS.OPENAI,
'mistral': SECRET_KEYS.MISTRALAI,
'google': SECRET_KEYS.MAKERSUITE,
'vertexai': SECRET_KEYS.VERTEXAI,
'anthropic': SECRET_KEYS.CLAUDE,
};

View File

@@ -22,6 +22,7 @@
<option value="cohere">Cohere</option>
<option value="custom" data-i18n="Custom (OpenAI-compatible)">Custom (OpenAI-compatible)</option>
<option value="google">Google AI Studio</option>
<option value="vertexai">Google Vertex AI</option>
<option value="groq">Groq</option>
<option value="koboldcpp">KoboldCpp</option>
<option value="llamacpp">llama.cpp</option>
@@ -111,6 +112,12 @@
<option data-type="google" value="gemini-1.5-flash-8b-exp-0827">gemini-1.5-flash-8b-exp-0827</option>
<option data-type="google" value="learnlm-2.0-flash-experimental">learnlm-2.0-flash-experimental</option>
<option data-type="google" value="learnlm-1.5-pro-experimental">learnlm-1.5-pro-experimental</option>
<option data-type="vertexai" value="gemini-2.5-pro-preview-05-06">gemini-2.5-pro-preview-05-06</option>
<option data-type="vertexai" value="gemini-2.5-pro-preview-03-25">gemini-2.5-pro-preview-03-25</option>
<option data-type="vertexai" value="gemini-2.5-flash-preview-05-20">gemini-2.5-flash-preview-05-20</option>
<option data-type="vertexai" value="gemini-2.5-flash-preview-04-17">gemini-2.5-flash-preview-04-17</option>
<option data-type="vertexai" value="gemini-2.0-flash-001">gemini-2.0-flash-001</option>
<option data-type="vertexai" value="gemini-2.0-flash-lite-001">gemini-2.0-flash-lite-001</option>
<option data-type="groq" value="llama-3.2-11b-vision-preview">llama-3.2-11b-vision-preview</option>
<option data-type="groq" value="llama-3.2-90b-vision-preview">llama-3.2-90b-vision-preview</option>
<option data-type="groq" value="llava-v1.5-7b-4096-preview">llava-v1.5-7b-4096-preview</option>
@@ -170,7 +177,7 @@
<div data-type="ollama">
The model must be downloaded first! Do it with the <code>ollama pull</code> command or <a href="#" id="caption_ollama_pull">click here</a>.
</div>
<label data-type="openai,anthropic,google,mistral" class="checkbox_label flexBasis100p" for="caption_allow_reverse_proxy" title="Allow using reverse proxy if defined and valid.">
<label data-type="openai,anthropic,google,vertexai,mistral" class="checkbox_label flexBasis100p" for="caption_allow_reverse_proxy" title="Allow using reverse proxy if defined and valid.">
<input id="caption_allow_reverse_proxy" type="checkbox" class="checkbox">
<span data-i18n="Allow reverse proxy">Allow reverse proxy</span>
</label>

View File

@@ -15,7 +15,7 @@ import { createThumbnail, isValidUrl } from '../utils.js';
*/
export async function getMultimodalCaption(base64Img, prompt) {
const useReverseProxy =
(['openai', 'anthropic', 'google', 'mistral'].includes(extension_settings.caption.multimodal_api))
(['openai', 'anthropic', 'google', 'mistral', 'vertexai'].includes(extension_settings.caption.multimodal_api))
&& extension_settings.caption.allow_reverse_proxy
&& oai_settings.reverse_proxy
&& isValidUrl(oai_settings.reverse_proxy);
@@ -38,7 +38,8 @@ export async function getMultimodalCaption(base64Img, prompt) {
const isVllm = extension_settings.caption.multimodal_api === 'vllm';
const base64Bytes = base64Img.length * 0.75;
const compressionLimit = 2 * 1024 * 1024;
if ((['google', 'openrouter', 'mistral', 'groq'].includes(extension_settings.caption.multimodal_api) && base64Bytes > compressionLimit) || isOoba || isKoboldCpp) {
const thumbnailNeeded = ['google', 'openrouter', 'mistral', 'groq', 'vertexai'].includes(extension_settings.caption.multimodal_api);
if ((thumbnailNeeded && base64Bytes > compressionLimit) || isOoba || isKoboldCpp) {
const maxSide = 1024;
base64Img = await createThumbnail(base64Img, maxSide, maxSide, 'image/jpeg');
}
@@ -94,6 +95,7 @@ export async function getMultimodalCaption(base64Img, prompt) {
function getEndpointUrl() {
switch (extension_settings.caption.multimodal_api) {
case 'google':
case 'vertexai':
return '/api/google/caption-image';
case 'anthropic':
return '/api/anthropic/caption-image';
@@ -143,6 +145,10 @@ function throwIfInvalidModel(useReverseProxy) {
throw new Error('Google AI Studio API key is not set.');
}
if (extension_settings.caption.multimodal_api === 'vertexai' && !secret_state[SECRET_KEYS.VERTEXAI] && !useReverseProxy) {
throw new Error('Google Vertex AI API key is not set.');
}
if (extension_settings.caption.multimodal_api === 'mistral' && !secret_state[SECRET_KEYS.MISTRALAI] && !useReverseProxy) {
throw new Error('Mistral AI API key is not set.');
}