lobehub
diff --git a/‎locales/ar/models.json
Lines changed: 36 additions & 9 deletions b/‎locales/ar/models.json
Lines changed: 36 additions & 9 deletions
diff --git a/‎locales/bg-BG/models.json
Lines changed: 36 additions & 9 deletions b/‎locales/bg-BG/models.json
Lines changed: 36 additions & 9 deletions
diff --git a/‎locales/de-DE/models.json
Lines changed: 36 additions & 9 deletions b/‎locales/de-DE/models.json
Lines changed: 36 additions & 9 deletions
diff --git a/‎locales/en-US/models.json
Lines changed: 36 additions & 9 deletions b/‎locales/en-US/models.json
Lines changed: 36 additions & 9 deletions
@@ -203,24 +203,21 @@
   "Pro/Qwen/Qwen2.5-VL-7B-Instruct": {
     "description": "Qwen2.5-VL is the newest addition to the Qwen series, featuring enhanced visual comprehension capabilities. It can analyze text, charts, and layouts within images, comprehend long videos while capturing events. The model supports reasoning, tool manipulation, multi-format object localization, and structured output generation. It incorporates optimized dynamic resolution and frame rate training for video understanding, along with improved efficiency in its visual encoder."
   },
+  "Pro/THUDM/GLM-4.1V-9B-Thinking": {
+    "description": "GLM-4.1V-9B-Thinking is an open-source vision-language model (VLM) jointly released by Zhipu AI and Tsinghua University's KEG Lab, designed specifically for handling complex multimodal cognitive tasks. Based on the GLM-4-9B-0414 foundation model, it significantly enhances cross-modal reasoning ability and stability by introducing the Chain-of-Thought reasoning mechanism and employing reinforcement learning strategies."
+  },
   "Pro/THUDM/glm-4-9b-chat": {
     "description": "GLM-4-9B-Chat is the open-source version of the GLM-4 series pre-trained models launched by Zhipu AI. This model excels in semantics, mathematics, reasoning, code, and knowledge. In addition to supporting multi-turn dialogues, GLM-4-9B-Chat also features advanced capabilities such as web browsing, code execution, custom tool invocation (Function Call), and long-text reasoning. The model supports 26 languages, including Chinese, English, Japanese, Korean, and German. In multiple benchmark tests, GLM-4-9B-Chat has demonstrated excellent performance, such as in AlignBench-v2, MT-Bench, MMLU, and C-Eval. The model supports a maximum context length of 128K, making it suitable for academic research and commercial applications."
   },
   "Pro/deepseek-ai/DeepSeek-R1": {
     "description": "DeepSeek-R1 is a reinforcement learning (RL) driven inference model that addresses issues of repetitiveness and readability in models. Prior to RL, DeepSeek-R1 introduced cold start data to further optimize inference performance. It performs comparably to OpenAI-o1 in mathematical, coding, and reasoning tasks, and enhances overall effectiveness through carefully designed training methods."
   },
-  "Pro/deepseek-ai/DeepSeek-R1-0120": {
-    "description": "DeepSeek-R1 is a reinforcement learning (RL) driven reasoning model that addresses issues of repetition and readability. Before RL, it introduced cold-start data to further optimize reasoning performance. It performs comparably to OpenAI-o1 in mathematics, coding, and reasoning tasks and improves overall effectiveness through carefully designed training methods."
-  },
   "Pro/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B": {
     "description": "DeepSeek-R1-Distill-Qwen-7B is a model derived from Qwen2.5-Math-7B through knowledge distillation. It was fine-tuned using 800,000 carefully selected samples generated by DeepSeek-R1, demonstrating exceptional reasoning capabilities. The model achieves outstanding performance across multiple benchmarks, including 92.8% accuracy on MATH-500, a 55.5% pass rate on AIME 2024, and a score of 1189 on CodeForces, showcasing strong mathematical and programming abilities for a 7B-scale model."
   },
   "Pro/deepseek-ai/DeepSeek-V3": {
     "description": "DeepSeek-V3 is a mixed expert (MoE) language model with 671 billion parameters, utilizing multi-head latent attention (MLA) and the DeepSeekMoE architecture, combined with a load balancing strategy without auxiliary loss to optimize inference and training efficiency. Pre-trained on 14.8 trillion high-quality tokens and fine-tuned with supervision and reinforcement learning, DeepSeek-V3 outperforms other open-source models and approaches leading closed-source models."
   },
-  "Pro/deepseek-ai/DeepSeek-V3-1226": {
-    "description": "DeepSeek-V3 is a mixture of experts (MoE) language model with 671 billion parameters, utilizing multi-head latent attention (MLA) and the DeepSeekMoE architecture, combined with a load balancing strategy without auxiliary loss to optimize inference and training efficiency. Pre-trained on 14.8 trillion high-quality tokens and fine-tuned with supervised learning and reinforcement learning, DeepSeek-V3 outperforms other open-source models and approaches leading closed-source models in performance."
-  },
   "QwQ-32B-Preview": {
     "description": "QwQ-32B-Preview is an innovative natural language processing model capable of efficiently handling complex dialogue generation and context understanding tasks."
   },
@@ -383,6 +380,9 @@
   "THUDM/GLM-4-9B-0414": {
     "description": "GLM-4-9B-0414 is a small model in the GLM series, with 9 billion parameters. This model inherits the technical characteristics of the GLM-4-32B series while providing a more lightweight deployment option. Despite its smaller size, GLM-4-9B-0414 still demonstrates excellent capabilities in tasks such as code generation, web design, SVG graphics generation, and search-based writing."
   },
+  "THUDM/GLM-4.1V-9B-Thinking": {
+    "description": "GLM-4.1V-9B-Thinking is an open-source vision-language model (VLM) jointly released by Zhipu AI and Tsinghua University's KEG Lab, designed specifically for handling complex multimodal cognitive tasks. Based on the GLM-4-9B-0414 foundation model, it significantly enhances cross-modal reasoning ability and stability by introducing the Chain-of-Thought reasoning mechanism and employing reinforcement learning strategies."
+  },
   "THUDM/GLM-Z1-32B-0414": {
     "description": "GLM-Z1-32B-0414 is a reasoning model with deep thinking capabilities. This model is developed based on GLM-4-32B-0414 through cold start and extended reinforcement learning, with further training in mathematics, coding, and logic tasks. Compared to the base model, GLM-Z1-32B-0414 significantly enhances mathematical abilities and the capacity to solve complex tasks."
   },
@@ -539,6 +539,9 @@
   "anthropic/claude-sonnet-4": {
     "description": "Claude Sonnet 4 can generate near-instant responses or extended step-by-step reasoning, allowing users to clearly observe these processes. API users also have fine-grained control over the model's thinking time."
   },
+  "ascend-tribe/pangu-pro-moe": {
+    "description": "Pangu-Pro-MoE 72B-A16B is a sparse large language model with 72 billion parameters and 16 billion activated parameters. It is based on the Group Mixture of Experts (MoGE) architecture, which groups experts during the expert selection phase and constrains tokens to activate an equal number of experts within each group, achieving expert load balancing and significantly improving deployment efficiency on the Ascend platform."
+  },
   "aya": {
     "description": "Aya 23 is a multilingual model launched by Cohere, supporting 23 languages, facilitating diverse language applications."
   },
@@ -548,6 +551,9 @@
   "baichuan/baichuan2-13b-chat": {
     "description": "Baichuan-13B is an open-source, commercially usable large language model developed by Baichuan Intelligence, containing 13 billion parameters, achieving the best results in its size on authoritative Chinese and English benchmarks."
   },
+  "baidu/ERNIE-4.5-300B-A47B": {
+    "description": "ERNIE-4.5-300B-A47B is a large language model developed by Baidu based on a Mixture of Experts (MoE) architecture. The model has a total of 300 billion parameters, but only activates 47 billion parameters per token during inference, balancing powerful performance with computational efficiency. As a core model in the ERNIE 4.5 series, it demonstrates outstanding capabilities in text understanding, generation, reasoning, and programming tasks. The model employs an innovative multimodal heterogeneous MoE pretraining method, jointly training text and visual modalities to effectively enhance overall capabilities, especially excelling in instruction following and world knowledge retention."
+  },
   "c4ai-aya-expanse-32b": {
     "description": "Aya Expanse is a high-performance 32B multilingual model designed to challenge the performance of single-language models through innovations in instruction tuning, data arbitrage, preference training, and model merging. It supports 23 languages."
   },
@@ -1097,9 +1103,6 @@
   "gemini-2.5-pro": {
     "description": "Gemini 2.5 Pro is Google's most advanced reasoning model, capable of tackling complex problems in coding, mathematics, and STEM fields, as well as analyzing large datasets, codebases, and documents using long-context processing."
   },
-  "gemini-2.5-pro-exp-03-25": {
-    "description": "Gemini 2.5 Pro Experimental is Google's most advanced thinking model, capable of reasoning about complex problems in code, mathematics, and STEM fields, as well as analyzing large datasets, codebases, and documents using long context."
-  },
   "gemini-2.5-pro-preview-03-25": {
     "description": "Gemini 2.5 Pro Preview is Google's most advanced thinking model, capable of reasoning about complex problems in code, mathematics, and STEM fields, as well as analyzing large datasets, codebases, and documents using long-context analysis."
   },
@@ -1166,6 +1169,12 @@
   "glm-4-plus": {
     "description": "GLM-4-Plus, as a high-intelligence flagship, possesses strong capabilities for processing long texts and complex tasks, with overall performance improvements."
   },
+  "glm-4.1v-thinking-flash": {
+    "description": "The GLM-4.1V-Thinking series represents the most powerful vision-language models known at the 10B parameter scale, integrating state-of-the-art capabilities across various vision-language tasks such as video understanding, image question answering, academic problem solving, OCR text recognition, document and chart interpretation, GUI agents, front-end web coding, and grounding. Its performance in many tasks even surpasses that of Qwen2.5-VL-72B, which has over eight times the parameters. Leveraging advanced reinforcement learning techniques, the model masters Chain-of-Thought reasoning to improve answer accuracy and richness, significantly outperforming traditional non-thinking models in final results and interpretability."
+  },
+  "glm-4.1v-thinking-flashx": {
+    "description": "The GLM-4.1V-Thinking series represents the most powerful vision-language models known at the 10B parameter scale, integrating state-of-the-art capabilities across various vision-language tasks such as video understanding, image question answering, academic problem solving, OCR text recognition, document and chart interpretation, GUI agents, front-end web coding, and grounding. Its performance in many tasks even surpasses that of Qwen2.5-VL-72B, which has over eight times the parameters. Leveraging advanced reinforcement learning techniques, the model masters Chain-of-Thought reasoning to improve answer accuracy and richness, significantly outperforming traditional non-thinking models in final results and interpretability."
+  },
   "glm-4v": {
     "description": "GLM-4V provides strong image understanding and reasoning capabilities, supporting various visual tasks."
   },
@@ -1187,6 +1196,9 @@
   "glm-z1-flash": {
     "description": "The GLM-Z1 series possesses strong complex reasoning capabilities, excelling in logical reasoning, mathematics, programming, and more. The maximum context length is 32K."
   },
+  "glm-z1-flashx": {
+    "description": "High speed and low cost: Flash enhanced version with ultra-fast inference speed and improved concurrency support."
+  },
   "glm-zero-preview": {
     "description": "GLM-Zero-Preview possesses strong complex reasoning abilities, excelling in logical reasoning, mathematics, programming, and other fields."
   },
@@ -1238,6 +1250,9 @@
   "google/gemma-2b-it": {
     "description": "Gemma Instruct (2B) provides basic instruction processing capabilities, suitable for lightweight applications."
   },
+  "google/gemma-3-1b-it": {
+    "description": "Gemma 3 1B is an open-source language model from Google that sets new standards in efficiency and performance."
+  },
   "google/gemma-3-27b-it": {
     "description": "Gemma 3 27B is an open-source language model from Google that sets new standards in efficiency and performance."
   },
@@ -1373,6 +1388,9 @@
   "gryphe/mythomax-l2-13b": {
     "description": "MythoMax l2 13B is a language model that combines creativity and intelligence by merging multiple top models."
   },
+  "hunyuan-a13b": {
+    "description": "Hunyuan's first hybrid reasoning model, an upgraded version of hunyuan-standard-256K, with a total of 80 billion parameters and 13 billion activated parameters. The default mode is slow thinking, supporting fast and slow thinking mode switching via parameters or instructions, with the switch implemented by adding 'query' prefix or 'no_think'. Overall capabilities are comprehensively improved compared to the previous generation, especially in mathematics, science, long text comprehension, and agent abilities."
+  },
   "hunyuan-code": {
     "description": "The latest code generation model from Hunyuan, trained on a base model with 200B high-quality code data, iteratively trained for six months with high-quality SFT data, increasing the context window length to 8K. It ranks among the top in automatic evaluation metrics for code generation across five major programming languages, and performs in the first tier for comprehensive human quality assessments across ten aspects of coding tasks."
   },
@@ -1424,6 +1442,9 @@
   "hunyuan-t1-vision": {
     "description": "Hunyuan is a multimodal deep thinking model supporting native multimodal chain-of-thought reasoning, excelling in various image reasoning scenarios and significantly outperforming fast-thinking models on science problems."
   },
+  "hunyuan-t1-vision-20250619": {
+    "description": "The latest Hunyuan t1-vision multimodal deep thinking model supports native long Chain-of-Thought reasoning across modalities, comprehensively improving over the previous default version."
+  },
   "hunyuan-turbo": {
     "description": "The preview version of the next-generation Hunyuan large language model, featuring a brand-new mixed expert model (MoE) structure, which offers faster inference efficiency and stronger performance compared to Hunyuan Pro."
   },
@@ -1454,6 +1475,12 @@
   "hunyuan-turbos-role-plus": {
     "description": "The latest Hunyuan role-playing model, officially fine-tuned and trained by Hunyuan. It is further trained on role-playing scenario datasets based on the Hunyuan model, delivering better foundational performance in role-playing contexts."
   },
+  "hunyuan-turbos-vision": {
+    "description": "This model is designed for image-text understanding scenarios and is based on Hunyuan's latest turbos architecture. It is a next-generation flagship vision-language model focusing on image-text understanding tasks, including image-based entity recognition, knowledge Q&A, copywriting, and photo-based problem solving, with comprehensive improvements over the previous generation."
+  },
+  "hunyuan-turbos-vision-20250619": {
+    "description": "The latest Hunyuan turbos-vision flagship vision-language model offers comprehensive improvements over the previous default version in image-text understanding tasks, including image-based entity recognition, knowledge Q&A, copywriting, and photo-based problem solving."
+  },
   "hunyuan-vision": {
     "description": "The latest multimodal model from Hunyuan, supporting image + text input to generate textual content."
   },