diff --git a/src/ts/model/modellist.ts b/src/ts/model/modellist.ts
index 2a01248190..3f55e67094 100644
--- a/src/ts/model/modellist.ts
+++ b/src/ts/model/modellist.ts
@@ -13,6 +13,7 @@ import {
 import { OpenAIModels } from './providers/openai'
 import { AnthropicModels } from './providers/anthropic'
 import { GoogleModels } from './providers/google'
+import { IonetModels } from './providers/ionet'
 import { fetchNative } from "../globalApi.svelte"
 import { DBState } from "../stores.svelte"
 import { customProviderStore, pluginV2 } from "../plugins/plugins.svelte"
@@ -44,6 +45,7 @@ function makeDeepInfraModels(id:string[]):LLMModel[]{
 export const LLMModels: LLMModel[] = [
     ...OpenAIModels,
     ...AnthropicModels,
+    ...IonetModels,
     // AWS Bedrock Claude models
     {
         name: "Claude 4.6 Opus v1",
diff --git a/src/ts/model/providers/ionet.ts b/src/ts/model/providers/ionet.ts
new file mode 100644
index 0000000000..a72c096fe9
--- /dev/null
+++ b/src/ts/model/providers/ionet.ts
@@ -0,0 +1,84 @@
+import { LLMFlags, LLMFormat, LLMProvider, LLMTokenizer, type LLMModel } from '../types'
+
+// IO Intelligence (io.net) — OpenAI-compatible Chat Completions gateway.
+// Endpoint: https://api.intelligence.io.solutions/api/v1/chat/completions
+// Model ids are org/name style (e.g. meta-llama/Llama-3.3-70B-Instruct), so
+// internalID carries the wire id and id stays a friendly unique key.
+// API keys: https://cloud.io.net
+
+export const IonetModels: LLMModel[] = [
+    {
+        id: 'ionet/glm-5.3',
+        internalID: 'zai-org/GLM-5.3',
+        name: 'GLM 5.3 (IO Intelligence)',
+        provider: LLMProvider.Ionet,
+        format: LLMFormat.OpenAICompatible,
+        flags: [LLMFlags.hasStreaming, LLMFlags.hasFullSystemPrompt],
+        parameters: ['frequency_penalty', 'presence_penalty', 'temperature', 'top_p'],
+        tokenizer: LLMTokenizer.GLM5,
+        endpoint: 'https://api.intelligence.io.solutions/api/v1/chat/completions',
+        keyIdentifier: 'ionet',
+        recommended: true
+    },
+    {
+        id: 'ionet/deepseek-v4.1-flash',
+        internalID: 'deepseek-ai/DeepSeek-V4.1-Flash',
+        name: 'DeepSeek V4.1 Flash (IO Intelligence)',
+        provider: LLMProvider.Ionet,
+        format: LLMFormat.OpenAICompatible,
+        flags: [LLMFlags.hasStreaming, LLMFlags.hasFullSystemPrompt],
+        parameters: ['frequency_penalty', 'presence_penalty', 'temperature', 'top_p'],
+        tokenizer: LLMTokenizer.DeepSeekV4,
+        endpoint: 'https://api.intelligence.io.solutions/api/v1/chat/completions',
+        keyIdentifier: 'ionet',
+        recommended: true
+    },
+    {
+        id: 'ionet/kimi-k3',
+        internalID: 'moonshotai/Kimi-K3',
+        name: 'Kimi K3 (IO Intelligence)',
+        provider: LLMProvider.Ionet,
+        format: LLMFormat.OpenAICompatible,
+        flags: [LLMFlags.hasStreaming, LLMFlags.hasFullSystemPrompt, LLMFlags.hasImageInput],
+        parameters: ['frequency_penalty', 'presence_penalty', 'temperature', 'top_p'],
+        tokenizer: LLMTokenizer.Unknown,
+        endpoint: 'https://api.intelligence.io.solutions/api/v1/chat/completions',
+        keyIdentifier: 'ionet'
+    },
+    {
+        id: 'ionet/qwen3.8-27b',
+        internalID: 'Qwen/Qwen3.8-27B',
+        name: 'Qwen3.8 27B (IO Intelligence)',
+        provider: LLMProvider.Ionet,
+        format: LLMFormat.OpenAICompatible,
+        flags: [LLMFlags.hasStreaming, LLMFlags.hasFullSystemPrompt, LLMFlags.hasImageInput],
+        parameters: ['frequency_penalty', 'presence_penalty', 'temperature', 'top_p'],
+        tokenizer: LLMTokenizer.Unknown,
+        endpoint: 'https://api.intelligence.io.solutions/api/v1/chat/completions',
+        keyIdentifier: 'ionet'
+    },
+    {
+        id: 'ionet/deepseek-r1-0528',
+        internalID: 'deepseek-ai/DeepSeek-R1-0528',
+        name: 'DeepSeek R1 0528 (IO Intelligence)',
+        provider: LLMProvider.Ionet,
+        format: LLMFormat.OpenAICompatible,
+        flags: [LLMFlags.hasStreaming, LLMFlags.hasFullSystemPrompt],
+        parameters: ['frequency_penalty', 'presence_penalty', 'temperature', 'top_p'],
+        tokenizer: LLMTokenizer.DeepSeek,
+        endpoint: 'https://api.intelligence.io.solutions/api/v1/chat/completions',
+        keyIdentifier: 'ionet'
+    },
+    {
+        id: 'ionet/llama-3.3-70b',
+        internalID: 'meta-llama/Llama-3.3-70B-Instruct',
+        name: 'Llama 3.3 70B (IO Intelligence)',
+        provider: LLMProvider.Ionet,
+        format: LLMFormat.OpenAICompatible,
+        flags: [LLMFlags.hasStreaming, LLMFlags.hasFullSystemPrompt],
+        parameters: ['frequency_penalty', 'presence_penalty', 'temperature', 'top_p'],
+        tokenizer: LLMTokenizer.Llama3,
+        endpoint: 'https://api.intelligence.io.solutions/api/v1/chat/completions',
+        keyIdentifier: 'ionet'
+    }
+]
diff --git a/src/ts/model/types.ts b/src/ts/model/types.ts
index bda9e9d376..6e1feff85a 100644
--- a/src/ts/model/types.ts
+++ b/src/ts/model/types.ts
@@ -48,7 +48,8 @@ export const LLMProvider = {
     DeepInfra: 13,
     Echo: 14,
     NanoGPT: 15,
-    Ollama: 16
+    Ollama: 16,
+    Ionet: 17
 } as const;
 export type LLMProvider = (typeof LLMProvider)[keyof typeof LLMProvider];
 
@@ -134,7 +135,8 @@ export const ProviderNames = new Map<LLMProvider, string>([
     [LLMProvider.DeepInfra, 'DeepInfra'],
     [LLMProvider.Echo, 'For Developer'],
     [LLMProvider.NanoGPT, 'NanoGPT'],
-    [LLMProvider.Ollama, 'Ollama']
+    [LLMProvider.Ollama, 'Ollama'],
+    [LLMProvider.Ionet, 'IO Intelligence']
 ])
 
 export const OpenAIParameters:LLMParameter[] = ['temperature', 'top_p', 'frequency_penalty', 'presence_penalty']
diff --git a/src/ts/tokenizer.ts b/src/ts/tokenizer.ts
index dbb7399403..6f23c4ffd6 100644
--- a/src/ts/tokenizer.ts
+++ b/src/ts/tokenizer.ts
@@ -196,6 +196,8 @@ export async function encode(data:string):Promise<(number[]|Uint32Array|Int32Arr
             result = await tokenizeWebTokenizers(data, 'GLM4');
         } else if(modelInfo.tokenizer === LLMTokenizer.GLM5){
             result = await tokenizeWebTokenizers(data, 'GLM5');
+        } else if(modelInfo.tokenizer === LLMTokenizer.Llama3){
+            result = await tokenizeWebTokenizers(data, 'llama3');
         } else if(modelInfo.tokenizer === LLMTokenizer.Cohere){
             result = await tokenizeWebTokenizers(data, 'cohere');
         } else {