From 43caa0c6d44f9b879f0636fc5e828ac3b31467f7 Mon Sep 17 00:00:00 2001
From: Gness Erquint <gness.na@gmail.com>
Date: Fri, 11 Apr 2025 11:09:10 +0300
Subject: [PATCH 01/12] Enable image inlining for visual models when connected
 to Mistral AI Le Platforme.

---
 public/index.html        | 2 +-
 public/scripts/openai.js | 4 +++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/public/index.html b/public/index.html
index d17b1041a..3286bf07f 100644
--- a/public/index.html
+++ b/public/index.html
@@ -1975,7 +1975,7 @@
                                             <span data-i18n="enable_functions_desc_3">Can be utilized by various extensions to provide additional functionality.</span>
                                         </div>
                                     </div>
-                                    <div class="range-block" data-source="openai,openrouter,makersuite,claude,custom,01ai">
+                                    <div class="range-block" data-source="openai,openrouter,mistralai,makersuite,claude,custom,01ai">
                                         <label for="openai_image_inlining" class="checkbox_label flexWrap widthFreeExpand">
                                             <input id="openai_image_inlining" type="checkbox" />
                                             <span data-i18n="Send inline images">Send inline images</span>
diff --git a/public/scripts/openai.js b/public/scripts/openai.js
index eb6051ff3..4fce97258 100644
--- a/public/scripts/openai.js
+++ b/public/scripts/openai.js
@@ -5006,7 +5006,9 @@ export function isImageInliningSupported() {
         'o1-2024-12-17',
         'chatgpt-4o-latest',
         'yi-vision',
-        'pixtral-latest',
+        'mistral-large-pixtral-2411',
+        'mistral-small-2503',
+        'mistral-small-latest',
         'pixtral-12b-latest',
         'pixtral-12b',
         'pixtral-12b-2409',

From c5ebe4b4b1fc38607f64a8058af2a528ecf56b82 Mon Sep 17 00:00:00 2001
From: Kristian Schlikow <Dakraid@users.noreply.github.com>
Date: Fri, 11 Apr 2025 12:09:13 +0000
Subject: [PATCH 02/12] Update the model filter for FAL.AI and update the list
 of providers for OpenRouter

---
 public/scripts/textgen-models.js  | 116 ++++++++++++++++--------------
 src/endpoints/stable-diffusion.js |   3 +-
 2 files changed, 63 insertions(+), 56 deletions(-)

diff --git a/public/scripts/textgen-models.js b/public/scripts/textgen-models.js
index fd70ffa6f..eb91aefef 100644
--- a/public/scripts/textgen-models.js
+++ b/public/scripts/textgen-models.js
@@ -23,61 +23,67 @@ export let openRouterModels = [];
  * @type {string[]}
  */
 const OPENROUTER_PROVIDERS = [
-    'OpenAI',
-    'Anthropic',
-    'Google',
-    'Google AI Studio',
-    'Amazon Bedrock',
-    'Groq',
-    'SambaNova',
-    'Cohere',
-    'Mistral',
-    'Together',
-    'Together 2',
-    'Fireworks',
-    'DeepInfra',
-    'Lepton',
-    'Novita',
-    'Avian',
-    'Lambda',
-    'Azure',
-    'Modal',
-    'AnyScale',
-    'Replicate',
-    'Perplexity',
-    'Recursal',
-    'OctoAI',
-    'DeepSeek',
-    'Infermatic',
-    'AI21',
-    'Featherless',
-    'Inflection',
-    'xAI',
-    'Cloudflare',
-    'SF Compute',
-    'Minimax',
-    'Nineteen',
-    'Liquid',
-    'InferenceNet',
-    'Friendli',
-    'AionLabs',
-    'Alibaba',
-    'Nebius',
-    'Chutes',
-    'Kluster',
-    'Crusoe',
-    'Targon',
-    'Ubicloud',
-    'Parasail',
-    '01.AI',
-    'HuggingFace',
-    'Mancer',
-    'Mancer 2',
-    'Hyperbolic',
-    'Hyperbolic 2',
-    'Lynn 2',
-    'Lynn',
-    'Reflection',
+
+    "OpenAI",
+    "Anthropic",
+    "Google",
+    "Google AI Studio",
+    "Amazon Bedrock",
+    "Groq",
+    "SambaNova",
+    "Cohere",
+    "Mistral",
+    "Together",
+    "Together 2",
+    "Fireworks",
+    "DeepInfra",
+    "Lepton",
+    "Novita",
+    "Avian",
+    "Lambda",
+    "Azure",
+    "Modal",
+    "AnyScale",
+    "Replicate",
+    "Perplexity",
+    "Recursal",
+    "OctoAI",
+    "DeepSeek",
+    "Infermatic",
+    "AI21",
+    "Featherless",
+    "Inflection",
+    "xAI",
+    "Cloudflare",
+    "SF Compute",
+    "Minimax",
+    "Nineteen",
+    "Liquid",
+    "Stealth",
+    "NCompass",
+    "InferenceNet",
+    "Friendli",
+    "AionLabs",
+    "Alibaba",
+    "Nebius",
+    "Chutes",
+    "Kluster",
+    "Crusoe",
+    "Targon",
+    "Ubicloud",
+    "Parasail",
+    "Phala",
+    "Cent-ML",
+    "Venice",
+    "01.AI",
+    "HuggingFace",
+    "Mancer",
+    "Mancer 2",
+    "Hyperbolic",
+    "Hyperbolic 2",
+    "Lynn 2",
+    "Lynn",
+    "Reflection"
 ];
 
 export async function loadOllamaModels(data) {
diff --git a/src/endpoints/stable-diffusion.js b/src/endpoints/stable-diffusion.js
index ee3944cb4..664c59399 100644
--- a/src/endpoints/stable-diffusion.js
+++ b/src/endpoints/stable-diffusion.js
@@ -1166,7 +1166,8 @@ falai.post('/models', async (_request, response) => {
         const models = data
             .filter(x => !x.title.toLowerCase().includes('inpainting') &&
                 !x.title.toLowerCase().includes('control') &&
-                !x.title.toLowerCase().includes('upscale'))
+                !x.title.toLowerCase().includes('upscale') &&
+                !x.title.toLowerCase().includes('lora'))
             .sort((a, b) => a.title.localeCompare(b.title))
             .map(x => ({ value: x.modelUrl.split('fal-ai/')[1], text: x.title }));
         return response.send(models);

From f238e5873115a0689a2993d027e735724002ee43 Mon Sep 17 00:00:00 2001
From: Kristian Schlikow <Dakraid@users.noreply.github.com>
Date: Fri, 11 Apr 2025 12:12:14 +0000
Subject: [PATCH 03/12] Remove unnecessary new line

---
 public/scripts/textgen-models.js | 1 -
 1 file changed, 1 deletion(-)

diff --git a/public/scripts/textgen-models.js b/public/scripts/textgen-models.js
index eb91aefef..61db3224f 100644
--- a/public/scripts/textgen-models.js
+++ b/public/scripts/textgen-models.js
@@ -23,7 +23,6 @@ export let openRouterModels = [];
  * @type {string[]}
  */
 const OPENROUTER_PROVIDERS = [
-
     "OpenAI",
     "Anthropic",
     "Google",

From e062a33f56b9b78a207f7171fd79bb3d264bbffb Mon Sep 17 00:00:00 2001
From: Kristian Schlikow <Dakraid@users.noreply.github.com>
Date: Fri, 11 Apr 2025 12:14:15 +0000
Subject: [PATCH 04/12] Use the proper quotes

---
 public/scripts/textgen-models.js | 120 +++++++++++++++----------------
 1 file changed, 60 insertions(+), 60 deletions(-)

diff --git a/public/scripts/textgen-models.js b/public/scripts/textgen-models.js
index 61db3224f..678c22982 100644
--- a/public/scripts/textgen-models.js
+++ b/public/scripts/textgen-models.js
@@ -23,66 +23,66 @@ export let openRouterModels = [];
  * @type {string[]}
  */
 const OPENROUTER_PROVIDERS = [
-    "OpenAI",
-    "Anthropic",
-    "Google",
-    "Google AI Studio",
-    "Amazon Bedrock",
-    "Groq",
-    "SambaNova",
-    "Cohere",
-    "Mistral",
-    "Together",
-    "Together 2",
-    "Fireworks",
-    "DeepInfra",
-    "Lepton",
-    "Novita",
-    "Avian",
-    "Lambda",
-    "Azure",
-    "Modal",
-    "AnyScale",
-    "Replicate",
-    "Perplexity",
-    "Recursal",
-    "OctoAI",
-    "DeepSeek",
-    "Infermatic",
-    "AI21",
-    "Featherless",
-    "Inflection",
-    "xAI",
-    "Cloudflare",
-    "SF Compute",
-    "Minimax",
-    "Nineteen",
-    "Liquid",
-    "Stealth",
-    "NCompass",
-    "InferenceNet",
-    "Friendli",
-    "AionLabs",
-    "Alibaba",
-    "Nebius",
-    "Chutes",
-    "Kluster",
-    "Crusoe",
-    "Targon",
-    "Ubicloud",
-    "Parasail",
-    "Phala",
-    "Cent-ML",
-    "Venice",
-    "01.AI",
-    "HuggingFace",
-    "Mancer",
-    "Mancer 2",
-    "Hyperbolic",
-    "Hyperbolic 2",
-    "Lynn 2",
-    "Lynn",
-    "Reflection"
+    'OpenAI',
+    'Anthropic',
+    'Google',
+    'Google AI Studio',
+    'Amazon Bedrock',
+    'Groq',
+    'SambaNova',
+    'Cohere',
+    'Mistral',
+    'Together',
+    'Together 2',
+    'Fireworks',
+    'DeepInfra',
+    'Lepton',
+    'Novita',
+    'Avian',
+    'Lambda',
+    'Azure',
+    'Modal',
+    'AnyScale',
+    'Replicate',
+    'Perplexity',
+    'Recursal',
+    'OctoAI',
+    'DeepSeek',
+    'Infermatic',
+    'AI21',
+    'Featherless',
+    'Inflection',
+    'xAI',
+    'Cloudflare',
+    'SF Compute',
+    'Minimax',
+    'Nineteen',
+    'Liquid',
+    'Stealth',
+    'NCompass',
+    'InferenceNet',
+    'Friendli',
+    'AionLabs',
+    'Alibaba',
+    'Nebius',
+    'Chutes',
+    'Kluster',
+    'Crusoe',
+    'Targon',
+    'Ubicloud',
+    'Parasail',
+    'Phala',
+    'Cent-ML',
+    'Venice',
+    '01.AI',
+    'HuggingFace',
+    'Mancer',
+    'Mancer 2',
+    'Hyperbolic',
+    'Hyperbolic 2',
+    'Lynn 2',
+    'Lynn',
+    'Reflection'
 ];
 
 export async function loadOllamaModels(data) {

From 797de862cfe91cc756fa31ec9e917bd0293a2103 Mon Sep 17 00:00:00 2001
From: Cohee <18619528+Cohee1207@users.noreply.github.com>
Date: Fri, 11 Apr 2025 15:15:29 +0300
Subject: [PATCH 05/12] Fix linter

---
 public/scripts/textgen-models.js | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/public/scripts/textgen-models.js b/public/scripts/textgen-models.js
index 678c22982..1aa6c8723 100644
--- a/public/scripts/textgen-models.js
+++ b/public/scripts/textgen-models.js
@@ -82,7 +82,7 @@ const OPENROUTER_PROVIDERS = [
     'Hyperbolic 2',
     'Lynn 2',
     'Lynn',
-    'Reflection'
+    'Reflection',
 ];
 
 export async function loadOllamaModels(data) {

From 0c4b0cfb0377fc55e591db3ef39a5e9980ee7a07 Mon Sep 17 00:00:00 2001
From: Gness Erquint <gness.na@gmail.com>
Date: Fri, 11 Apr 2025 17:20:39 +0300
Subject: [PATCH 06/12] Set correct Mistral AI token context limits.

---
 public/scripts/openai.js | 21 +++++++++------------
 1 file changed, 9 insertions(+), 12 deletions(-)

diff --git a/public/scripts/openai.js b/public/scripts/openai.js
index 4fce97258..9d01a4a13 100644
--- a/public/scripts/openai.js
+++ b/public/scripts/openai.js
@@ -4430,20 +4430,17 @@ async function onModelChange() {
     if (oai_settings.chat_completion_source === chat_completion_sources.MISTRALAI) {
         if (oai_settings.max_context_unlocked) {
             $('#openai_max_context').attr('max', unlocked_max);
-        } else if (['codestral-latest', 'codestral-mamba-2407', 'codestral-2411-rc5', 'codestral-2412', 'codestral-2501'].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', max_256k);
-        } else if (['mistral-large-2407', 'mistral-large-2411', 'mistral-large-pixtral-2411', 'mistral-large-latest'].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', max_128k);
-        } else if (oai_settings.mistralai_model.includes('mistral-nemo')) {
-            $('#openai_max_context').attr('max', max_128k);
-        } else if (oai_settings.mistralai_model.includes('mixtral-8x22b')) {
-            $('#openai_max_context').attr('max', max_64k);
-        } else if (oai_settings.mistralai_model.includes('pixtral')) {
-            $('#openai_max_context').attr('max', max_128k);
+        } else if (["codestral-2411-rc5", "codestral-2412", "codestral-2501", "codestral-latest", "codestral-mamba-2407", "codestral-mamba-latest", "open-codestral-mamba"].includes(oai_settings.mistralai_model)) {
+            $('#openai_max_context').attr('max', 262144);
+        } else if (["ministral-3b-2410", "ministral-3b-latest", "ministral-8b-2410", "ministral-8b-latest", "mistral-large-2407", "mistral-large-2411", "mistral-large-latest", "mistral-large-pixtral-2411", "mistral-tiny-2407", "mistral-tiny-latest", "open-mistral-nemo", "open-mistral-nemo-2407", "pixtral-12b", "pixtral-12b-2409", "pixtral-12b-latest", "pixtral-large-2411", "pixtral-large-latest"].includes(oai_settings.mistralai_model)) {
+            $('#openai_max_context').attr('max', 131072);
+        } else if (["open-mixtral-8x22b", "open-mixtral-8x22b-2404"].includes(oai_settings.mistralai_model)) {
+            $('#openai_max_context').attr('max', 65536);
         } else if (oai_settings.mistralai_model.includes('ministral')) {
-            $('#openai_max_context').attr('max', max_32k);
+        } else if (["codestral-2405", "mistral-embed", "mistral-large-2402", "mistral-medium", "mistral-medium-2312", "mistral-medium-latest", "mistral-moderation-2411", "mistral-moderation-latest", "mistral-ocr-2503", "mistral-ocr-latest", "mistral-saba-2502", "mistral-saba-latest", "mistral-small", "mistral-small-2312", "mistral-small-2402", "mistral-small-2409", "mistral-small-2501", "mistral-small-2503", "mistral-small-latest", "mistral-tiny", "mistral-tiny-2312", "open-mistral-7b", "open-mixtral-8x7b"].includes(oai_settings.mistralai_model)) {
+            $('#openai_max_context').attr('max', 32768);
         } else {
-            $('#openai_max_context').attr('max', max_32k);
+            $('#openai_max_context').attr('max', 32768);
         }
         oai_settings.openai_max_context = Math.min(oai_settings.openai_max_context, Number($('#openai_max_context').attr('max')));
         $('#openai_max_context').val(oai_settings.openai_max_context).trigger('input');

From 2d77fb3e3094a1440640de1f0381d1e87edb0939 Mon Sep 17 00:00:00 2001
From: Gness Erquint <gness.na@gmail.com>
Date: Fri, 11 Apr 2025 17:40:53 +0300
Subject: [PATCH 07/12] Powers of 2 for token context limits. No -1 offset.

---
 public/scripts/openai.js | 30 +++++++++++++++---------------
 1 file changed, 15 insertions(+), 15 deletions(-)

diff --git a/public/scripts/openai.js b/public/scripts/openai.js
index 9d01a4a13..55db12f5b 100644
--- a/public/scripts/openai.js
+++ b/public/scripts/openai.js
@@ -117,17 +117,17 @@ const default_bias_presets = {
     ],
 };
 
-const max_2k = 2047;
-const max_4k = 4095;
-const max_8k = 8191;
-const max_16k = 16383;
-const max_32k = 32767;
+const max_2k = 2048;
+const max_4k = 4096;
+const max_8k = 8192;
+const max_16k = 16384;
+const max_32k = 32768;
 const max_64k = 65535;
-const max_128k = 128 * 1000;
-const max_200k = 200 * 1000;
-const max_256k = 256 * 1000;
-const max_1mil = 1000 * 1000;
-const max_2mil = 2000 * 1000;
+const max_128k = 131072;
+const max_200k = 204800;
+const max_256k = 262144;
+const max_1mil = 1048576;
+const max_2mil = 2097152;
 const scale_max = 8191;
 const claude_max = 9000; // We have a proper tokenizer, so theoretically could be larger (up to 9k)
 const claude_100k_max = 99000;
@@ -4431,16 +4431,16 @@ async function onModelChange() {
         if (oai_settings.max_context_unlocked) {
             $('#openai_max_context').attr('max', unlocked_max);
         } else if (["codestral-2411-rc5", "codestral-2412", "codestral-2501", "codestral-latest", "codestral-mamba-2407", "codestral-mamba-latest", "open-codestral-mamba"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', 262144);
+            $('#openai_max_context').attr('max', max_256k);
         } else if (["ministral-3b-2410", "ministral-3b-latest", "ministral-8b-2410", "ministral-8b-latest", "mistral-large-2407", "mistral-large-2411", "mistral-large-latest", "mistral-large-pixtral-2411", "mistral-tiny-2407", "mistral-tiny-latest", "open-mistral-nemo", "open-mistral-nemo-2407", "pixtral-12b", "pixtral-12b-2409", "pixtral-12b-latest", "pixtral-large-2411", "pixtral-large-latest"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', 131072);
+            $('#openai_max_context').attr('max', max_128k);
         } else if (["open-mixtral-8x22b", "open-mixtral-8x22b-2404"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', 65536);
+            $('#openai_max_context').attr('max', max_64k);
         } else if (oai_settings.mistralai_model.includes('ministral')) {
         } else if (["codestral-2405", "mistral-embed", "mistral-large-2402", "mistral-medium", "mistral-medium-2312", "mistral-medium-latest", "mistral-moderation-2411", "mistral-moderation-latest", "mistral-ocr-2503", "mistral-ocr-latest", "mistral-saba-2502", "mistral-saba-latest", "mistral-small", "mistral-small-2312", "mistral-small-2402", "mistral-small-2409", "mistral-small-2501", "mistral-small-2503", "mistral-small-latest", "mistral-tiny", "mistral-tiny-2312", "open-mistral-7b", "open-mixtral-8x7b"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', 32768);
+            $('#openai_max_context').attr('max', max_32k);
         } else {
-            $('#openai_max_context').attr('max', 32768);
+            $('#openai_max_context').attr('max', max_32k);
         }
         oai_settings.openai_max_context = Math.min(oai_settings.openai_max_context, Number($('#openai_max_context').attr('max')));
         $('#openai_max_context').val(oai_settings.openai_max_context).trigger('input');

From 2040c43371a73c97e127bc47ec1b8905e4f16fc6 Mon Sep 17 00:00:00 2001
From: Gness Erquint <gness.na@gmail.com>
Date: Fri, 11 Apr 2025 17:58:39 +0300
Subject: [PATCH 08/12] Revert "Powers of 2 for token context limits. No -1
 offset."

This reverts commit 2d77fb3e3094a1440640de1f0381d1e87edb0939.
---
 public/scripts/openai.js | 30 +++++++++++++++---------------
 1 file changed, 15 insertions(+), 15 deletions(-)

diff --git a/public/scripts/openai.js b/public/scripts/openai.js
index 55db12f5b..9d01a4a13 100644
--- a/public/scripts/openai.js
+++ b/public/scripts/openai.js
@@ -117,17 +117,17 @@ const default_bias_presets = {
     ],
 };
 
-const max_2k = 2048;
-const max_4k = 4096;
-const max_8k = 8192;
-const max_16k = 16384;
-const max_32k = 32768;
+const max_2k = 2047;
+const max_4k = 4095;
+const max_8k = 8191;
+const max_16k = 16383;
+const max_32k = 32767;
 const max_64k = 65535;
-const max_128k = 131072;
-const max_200k = 204800;
-const max_256k = 262144;
-const max_1mil = 1048576;
-const max_2mil = 2097152;
+const max_128k = 128 * 1000;
+const max_200k = 200 * 1000;
+const max_256k = 256 * 1000;
+const max_1mil = 1000 * 1000;
+const max_2mil = 2000 * 1000;
 const scale_max = 8191;
 const claude_max = 9000; // We have a proper tokenizer, so theoretically could be larger (up to 9k)
 const claude_100k_max = 99000;
@@ -4431,16 +4431,16 @@ async function onModelChange() {
         if (oai_settings.max_context_unlocked) {
             $('#openai_max_context').attr('max', unlocked_max);
         } else if (["codestral-2411-rc5", "codestral-2412", "codestral-2501", "codestral-latest", "codestral-mamba-2407", "codestral-mamba-latest", "open-codestral-mamba"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', max_256k);
+            $('#openai_max_context').attr('max', 262144);
         } else if (["ministral-3b-2410", "ministral-3b-latest", "ministral-8b-2410", "ministral-8b-latest", "mistral-large-2407", "mistral-large-2411", "mistral-large-latest", "mistral-large-pixtral-2411", "mistral-tiny-2407", "mistral-tiny-latest", "open-mistral-nemo", "open-mistral-nemo-2407", "pixtral-12b", "pixtral-12b-2409", "pixtral-12b-latest", "pixtral-large-2411", "pixtral-large-latest"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', max_128k);
+            $('#openai_max_context').attr('max', 131072);
         } else if (["open-mixtral-8x22b", "open-mixtral-8x22b-2404"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', max_64k);
+            $('#openai_max_context').attr('max', 65536);
         } else if (oai_settings.mistralai_model.includes('ministral')) {
         } else if (["codestral-2405", "mistral-embed", "mistral-large-2402", "mistral-medium", "mistral-medium-2312", "mistral-medium-latest", "mistral-moderation-2411", "mistral-moderation-latest", "mistral-ocr-2503", "mistral-ocr-latest", "mistral-saba-2502", "mistral-saba-latest", "mistral-small", "mistral-small-2312", "mistral-small-2402", "mistral-small-2409", "mistral-small-2501", "mistral-small-2503", "mistral-small-latest", "mistral-tiny", "mistral-tiny-2312", "open-mistral-7b", "open-mixtral-8x7b"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', max_32k);
+            $('#openai_max_context').attr('max', 32768);
         } else {
-            $('#openai_max_context').attr('max', max_32k);
+            $('#openai_max_context').attr('max', 32768);
         }
         oai_settings.openai_max_context = Math.min(oai_settings.openai_max_context, Number($('#openai_max_context').attr('max')));
         $('#openai_max_context').val(oai_settings.openai_max_context).trigger('input');

From 1d2122b867c1cb6a4d28cf91e98c6ec05f639d4d Mon Sep 17 00:00:00 2001
From: Gness Erquint <gness.na@gmail.com>
Date: Fri, 11 Apr 2025 18:01:42 +0300
Subject: [PATCH 09/12] Correct editing mistake in "Set correct Mistral AI
 token context limits."

---
 public/scripts/openai.js | 1 -
 1 file changed, 1 deletion(-)

diff --git a/public/scripts/openai.js b/public/scripts/openai.js
index 9d01a4a13..cfa3f530b 100644
--- a/public/scripts/openai.js
+++ b/public/scripts/openai.js
@@ -4436,7 +4436,6 @@ async function onModelChange() {
             $('#openai_max_context').attr('max', 131072);
         } else if (["open-mixtral-8x22b", "open-mixtral-8x22b-2404"].includes(oai_settings.mistralai_model)) {
             $('#openai_max_context').attr('max', 65536);
-        } else if (oai_settings.mistralai_model.includes('ministral')) {
         } else if (["codestral-2405", "mistral-embed", "mistral-large-2402", "mistral-medium", "mistral-medium-2312", "mistral-medium-latest", "mistral-moderation-2411", "mistral-moderation-latest", "mistral-ocr-2503", "mistral-ocr-latest", "mistral-saba-2502", "mistral-saba-latest", "mistral-small", "mistral-small-2312", "mistral-small-2402", "mistral-small-2409", "mistral-small-2501", "mistral-small-2503", "mistral-small-latest", "mistral-tiny", "mistral-tiny-2312", "open-mistral-7b", "open-mixtral-8x7b"].includes(oai_settings.mistralai_model)) {
             $('#openai_max_context').attr('max', 32768);
         } else {

From 4736f533a5b75dc2f612b5fa9fe81f96c2465555 Mon Sep 17 00:00:00 2001
From: bmen25124 <bmen25124@gmail.com>
Date: Sat, 5 Apr 2025 16:10:42 +0300
Subject: [PATCH 10/12] Added proxy support to ChatCompletionService

---
 public/scripts/custom-request.js    | 10 +++++++---
 public/scripts/extensions/shared.js |  6 +++++-
 2 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/public/scripts/custom-request.js b/public/scripts/custom-request.js
index 5cd3f4827..69c1ab882 100644
--- a/public/scripts/custom-request.js
+++ b/public/scripts/custom-request.js
@@ -43,10 +43,12 @@ import EventSourceStream from './sse-stream.js';
  * @property {boolean?} [stream=false] - Whether to stream the response
  * @property {ChatCompletionMessage[]} messages - Array of chat messages
  * @property {string} [model] - Optional model name to use for completion
- * @property {string} chat_completion_source - Source provider for chat completion
+ * @property {string} chat_completion_source - Source provider
  * @property {number} max_tokens - Maximum number of tokens to generate
  * @property {number} [temperature] - Optional temperature parameter for response randomness
- * @property {string} [custom_url] - Optional custom URL for chat completion
+ * @property {string} [custom_url] - Optional custom URL
+ * @property {string} [reverse_proxy] - Optional reverse proxy URL
+ * @property {string} [proxy_password] - Optional proxy password
  */
 
 /** @typedef {Record<string, any> & ChatCompletionPayloadBase} ChatCompletionPayload */
@@ -387,7 +389,7 @@ export class ChatCompletionService {
      * @param {ChatCompletionPayload} custom
      * @returns {ChatCompletionPayload}
      */
-    static createRequestData({ stream = false, messages, model, chat_completion_source, max_tokens, temperature, custom_url, ...props }) {
+    static createRequestData({ stream = false, messages, model, chat_completion_source, max_tokens, temperature, custom_url, reverse_proxy, proxy_password, ...props }) {
         const payload = {
             ...props,
             stream,
@@ -397,6 +399,8 @@ export class ChatCompletionService {
             max_tokens,
             temperature,
             custom_url,
+            reverse_proxy,
+            proxy_password,
         };
 
         // Remove undefined values to avoid API errors
diff --git a/public/scripts/extensions/shared.js b/public/scripts/extensions/shared.js
index 3ac65b894..dfd4bc2c2 100644
--- a/public/scripts/extensions/shared.js
+++ b/public/scripts/extensions/shared.js
@@ -1,7 +1,7 @@
 import { CONNECT_API_MAP, getRequestHeaders } from '../../script.js';
 import { extension_settings, openThirdPartyExtensionMenu } from '../extensions.js';
 import { t } from '../i18n.js';
-import { oai_settings } from '../openai.js';
+import { oai_settings, proxies } from '../openai.js';
 import { SECRET_KEYS, secret_state } from '../secrets.js';
 import { textgen_types, textgenerationwebui_settings } from '../textgen-settings.js';
 import { getTokenCountAsync } from '../tokenizers.js';
@@ -326,6 +326,8 @@ export class ConnectionManagerRequestService {
                         throw new Error(`API type ${selectedApiMap.selected} does not support chat completions`);
                     }
 
+                    const proxyPreset = proxies.find((p) => p.name === profile.proxy);
+
                     const messages = Array.isArray(prompt) ? prompt : [{ role: 'user', content: prompt }];
                     return await context.ChatCompletionService.processRequest({
                         stream,
@@ -334,6 +336,8 @@ export class ConnectionManagerRequestService {
                         model: profile.model,
                         chat_completion_source: selectedApiMap.source,
                         custom_url: profile['api-url'],
+                        reverse_proxy: proxyPreset?.url,
+                        proxy_password: proxyPreset?.password,
                     }, {
                         presetName: includePreset ? profile.preset : undefined,
                     }, extractData, signal);

From fc5e0563bae0dc17113d8b3a657f8cdea2ae51b2 Mon Sep 17 00:00:00 2001
From: bmen25124 <bmen25124@gmail.com>
Date: Fri, 11 Apr 2025 19:07:00 +0300
Subject: [PATCH 11/12] Added ability to override request payload

---
 public/scripts/custom-request.js    | 6 ++++--
 public/scripts/extensions/shared.js | 5 ++++-
 2 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/public/scripts/custom-request.js b/public/scripts/custom-request.js
index 69c1ab882..888a09745 100644
--- a/public/scripts/custom-request.js
+++ b/public/scripts/custom-request.js
@@ -82,7 +82,6 @@ export class TextCompletionService {
      */
     static createRequestData({ stream = false, prompt, max_tokens, model, api_type, api_server, temperature, min_p, ...props }) {
         const payload = {
-            ...props,
             stream,
             prompt,
             max_tokens,
@@ -92,6 +91,7 @@ export class TextCompletionService {
             api_server: api_server ?? getTextGenServer(api_type),
             temperature,
             min_p,
+            ...props,
         };
 
         // Remove undefined values to avoid API errors
@@ -391,7 +391,6 @@ export class ChatCompletionService {
      */
     static createRequestData({ stream = false, messages, model, chat_completion_source, max_tokens, temperature, custom_url, reverse_proxy, proxy_password, ...props }) {
         const payload = {
-            ...props,
             stream,
             messages,
             model,
@@ -401,6 +400,9 @@ export class ChatCompletionService {
             custom_url,
             reverse_proxy,
             proxy_password,
+            use_makersuite_sysprompt: true,
+            claude_use_sysprompt: true,
+            ...props,
         };
 
         // Remove undefined values to avoid API errors
diff --git a/public/scripts/extensions/shared.js b/public/scripts/extensions/shared.js
index dfd4bc2c2..0568cbb24 100644
--- a/public/scripts/extensions/shared.js
+++ b/public/scripts/extensions/shared.js
@@ -306,9 +306,10 @@ export class ConnectionManagerRequestService {
      * @param {boolean?} [custom.includePreset=true]
      * @param {boolean?} [custom.includeInstruct=true]
      * @param {Partial<InstructSettings>?} [custom.instructSettings] Override instruct settings
+     * @param {Record<string, any>} [overridePayload] - Override payload for the request
      * @returns {Promise<import('../custom-request.js').ExtractedData | (() => AsyncGenerator<import('../custom-request.js').StreamResponse>)>} If not streaming, returns extracted data; if streaming, returns a function that creates an AsyncGenerator
      */
-    static async sendRequest(profileId, prompt, maxTokens, custom = this.defaultSendRequestParams) {
+    static async sendRequest(profileId, prompt, maxTokens, custom = this.defaultSendRequestParams, overridePayload = {}) {
         const { stream, signal, extractData, includePreset, includeInstruct, instructSettings } = { ...this.defaultSendRequestParams, ...custom };
 
         const context = SillyTavern.getContext();
@@ -338,6 +339,7 @@ export class ConnectionManagerRequestService {
                         custom_url: profile['api-url'],
                         reverse_proxy: proxyPreset?.url,
                         proxy_password: proxyPreset?.password,
+                        ...overridePayload,
                     }, {
                         presetName: includePreset ? profile.preset : undefined,
                     }, extractData, signal);
@@ -354,6 +356,7 @@ export class ConnectionManagerRequestService {
                         model: profile.model,
                         api_type: selectedApiMap.type,
                         api_server: profile['api-url'],
+                        ...overridePayload,
                     }, {
                         instructName: includeInstruct ? profile.instruct : undefined,
                         presetName: includePreset ? profile.preset : undefined,

From 1f27a39f2928230fef84294ef66d553d30ac2116 Mon Sep 17 00:00:00 2001
From: Cohee <18619528+Cohee1207@users.noreply.github.com>
Date: Fri, 11 Apr 2025 21:09:06 +0300
Subject: [PATCH 12/12] Refactor mistral max context

---
 public/scripts/openai.js | 91 +++++++++++++++++++++++++++++++++-------
 1 file changed, 76 insertions(+), 15 deletions(-)

diff --git a/public/scripts/openai.js b/public/scripts/openai.js
index cfa3f530b..82209297d 100644
--- a/public/scripts/openai.js
+++ b/public/scripts/openai.js
@@ -4165,6 +4165,80 @@ function getMaxContextWindowAI(value) {
     }
 }
 
+/**
+ * Get the maximum context size for the Mistral model
+ * @param {string} model Model identifier
+ * @param {boolean} isUnlocked Whether context limits are unlocked
+ * @returns {number} Maximum context size in tokens
+ */
+function getMistralMaxContext(model, isUnlocked) {
+    if (isUnlocked) {
+        return unlocked_max;
+    }
+
+    if (Array.isArray(model_list) && model_list.length > 0) {
+        const contextLength = model_list.find((record) => record.id === model)?.max_context_length;
+        if (contextLength) {
+            return contextLength;
+        }
+    }
+
+    const contextMap = {
+        'codestral-2411-rc5': 262144,
+        'codestral-2412': 262144,
+        'codestral-2501': 262144,
+        'codestral-latest': 262144,
+        'codestral-mamba-2407': 262144,
+        'codestral-mamba-latest': 262144,
+        'open-codestral-mamba': 262144,
+        'ministral-3b-2410': 131072,
+        'ministral-3b-latest': 131072,
+        'ministral-8b-2410': 131072,
+        'ministral-8b-latest': 131072,
+        'mistral-large-2407': 131072,
+        'mistral-large-2411': 131072,
+        'mistral-large-latest': 131072,
+        'mistral-large-pixtral-2411': 131072,
+        'mistral-tiny-2407': 131072,
+        'mistral-tiny-latest': 131072,
+        'open-mistral-nemo': 131072,
+        'open-mistral-nemo-2407': 131072,
+        'pixtral-12b': 131072,
+        'pixtral-12b-2409': 131072,
+        'pixtral-12b-latest': 131072,
+        'pixtral-large-2411': 131072,
+        'pixtral-large-latest': 131072,
+        'open-mixtral-8x22b': 65536,
+        'open-mixtral-8x22b-2404': 65536,
+        'codestral-2405': 32768,
+        'mistral-embed': 32768,
+        'mistral-large-2402': 32768,
+        'mistral-medium': 32768,
+        'mistral-medium-2312': 32768,
+        'mistral-medium-latest': 32768,
+        'mistral-moderation-2411': 32768,
+        'mistral-moderation-latest': 32768,
+        'mistral-ocr-2503': 32768,
+        'mistral-ocr-latest': 32768,
+        'mistral-saba-2502': 32768,
+        'mistral-saba-latest': 32768,
+        'mistral-small': 32768,
+        'mistral-small-2312': 32768,
+        'mistral-small-2402': 32768,
+        'mistral-small-2409': 32768,
+        'mistral-small-2501': 32768,
+        'mistral-small-2503': 32768,
+        'mistral-small-latest': 32768,
+        'mistral-tiny': 32768,
+        'mistral-tiny-2312': 32768,
+        'open-mistral-7b': 32768,
+        'open-mixtral-8x7b': 32768,
+    };
+
+    // Return context size if model found, otherwise default to 32k
+    return Object.entries(contextMap).find(([key]) => model.includes(key))?.[1] || 32768;
+}
+
 /**
  * Get the maximum context size for the Groq model
  * @param {string} model Model identifier
@@ -4428,23 +4502,10 @@ async function onModelChange() {
     }
 
     if (oai_settings.chat_completion_source === chat_completion_sources.MISTRALAI) {
-        if (oai_settings.max_context_unlocked) {
-            $('#openai_max_context').attr('max', unlocked_max);
-        } else if (["codestral-2411-rc5", "codestral-2412", "codestral-2501", "codestral-latest", "codestral-mamba-2407", "codestral-mamba-latest", "open-codestral-mamba"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', 262144);
-        } else if (["ministral-3b-2410", "ministral-3b-latest", "ministral-8b-2410", "ministral-8b-latest", "mistral-large-2407", "mistral-large-2411", "mistral-large-latest", "mistral-large-pixtral-2411", "mistral-tiny-2407", "mistral-tiny-latest", "open-mistral-nemo", "open-mistral-nemo-2407", "pixtral-12b", "pixtral-12b-2409", "pixtral-12b-latest", "pixtral-large-2411", "pixtral-large-latest"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', 131072);
-        } else if (["open-mixtral-8x22b", "open-mixtral-8x22b-2404"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', 65536);
-        } else if (["codestral-2405", "mistral-embed", "mistral-large-2402", "mistral-medium", "mistral-medium-2312", "mistral-medium-latest", "mistral-moderation-2411", "mistral-moderation-latest", "mistral-ocr-2503", "mistral-ocr-latest", "mistral-saba-2502", "mistral-saba-latest", "mistral-small", "mistral-small-2312", "mistral-small-2402", "mistral-small-2409", "mistral-small-2501", "mistral-small-2503", "mistral-small-latest", "mistral-tiny", "mistral-tiny-2312", "open-mistral-7b", "open-mixtral-8x7b"].includes(oai_settings.mistralai_model)) {
-            $('#openai_max_context').attr('max', 32768);
-        } else {
-            $('#openai_max_context').attr('max', 32768);
-        }
+        const maxContext = getMistralMaxContext(oai_settings.mistralai_model, oai_settings.max_context_unlocked);
+        $('#openai_max_context').attr('max', maxContext);
         oai_settings.openai_max_context = Math.min(oai_settings.openai_max_context, Number($('#openai_max_context').attr('max')));
         $('#openai_max_context').val(oai_settings.openai_max_context).trigger('input');
-
-        //mistral also caps temp at 1.0
         oai_settings.temp_openai = Math.min(claude_max_temp, oai_settings.temp_openai);
         $('#temp_openai').attr('max', claude_max_temp).val(oai_settings.temp_openai).trigger('input');
     }