feat(sync): add Novita AI model catalog sync

.github/workflows/sync-models.yml

@@ -84,6 +84,7 @@ jobs:
           VENICE_API_KEY: ${{ secrets.VENICE_API_KEY }}
           LLMGATEWAY_API_KEY: ${{ secrets.LLMGATEWAY_API_KEY }}
           MERGE_GATEWAY_API_KEY: ${{ secrets.MERGE_GATEWAY_API_KEY }}
+          NOVITA_API_KEY: ${{ secrets.NOVITA_API_KEY }}
           MISTRAL_API_KEY: ${{ secrets.MISTRAL_API_KEY }}
           KILO_API_KEY: ${{ secrets.KILO_API_KEY }}
           GOOGLE_API_KEY: ${{ secrets.GOOGLE_API_KEY }}

models/alibaba/qwen-mt-plus.toml

@@ -0,0 +1,19 @@
+name = "Qwen MT Plus"
+description = "Qwen translation model for multilingual conversion and localization"
+family = "qwen"
+release_date = "2025-09-03"
+last_updated = "2025-09-03"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = false
+structured_output = false
+open_weights = true
+
+[limit]
+context = 16_384
+output = 8_192
+
+[modalities]
+input = ["text"]
+output = ["text"]

models/alibaba/qwen2.5-72b-instruct.toml

@@ -0,0 +1,20 @@
+name = "Qwen2.5 72B Instruct"
+description = "Open-weight Qwen2.5 instruction model for multilingual chat and coding"
+family = "qwen"
+release_date = "2024-10-15"
+last_updated = "2024-10-15"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2024-04"
+open_weights = true
+
+[limit]
+context = 32_000
+output = 8_192
+
+[modalities]
+input = ["text"]
+output = ["text"]

models/alibaba/qwen2.5-7b-instruct.toml

@@ -0,0 +1,25 @@
+# https://huggingface.co/Qwen/Qwen2.5-7B-Instruct
+name = "Qwen2.5 7B Instruct"
+description = "Open Qwen instruction model for multilingual chat, coding, and structured responses"
+family = "qwen"
+release_date = "2024-09-19"
+last_updated = "2024-09-19"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+license = "Apache-2.0"
+
+[limit]
+context = 131_072
+output = 8_192
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen2.5-7B-Instruct"

models/alibaba/qwen3-235b-a22b-thinking-2507.toml

@@ -0,0 +1,24 @@
+# https://huggingface.co/Qwen/Qwen3-235B-A22B-Thinking-2507
+name = "Qwen3 235B A22B Thinking 2507"
+description = "Qwen reasoning model for deliberate problem solving, math, coding, and agentic workflows"
+family = "qwen"
+release_date = "2025-07-25"
+last_updated = "2025-07-25"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = true
+
+[limit]
+context = 131_072
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-235B-A22B-Thinking-2507"

models/alibaba/qwen3-omni-30b-a3b-instruct.toml

@@ -0,0 +1,20 @@
+name = "Qwen3 Omni 30B A3B Instruct"
+description = "Qwen3 omni instruction model for text, vision, audio, and multimodal tasks"
+family = "qwen"
+release_date = "2025-09-24"
+last_updated = "2025-09-24"
+attachment = true
+reasoning = false
+temperature = true
+knowledge = "2024-04"
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 65_536
+output = 16_384
+
+[modalities]
+input = ["text", "video", "audio", "image"]
+output = ["text", "audio"]

models/alibaba/qwen3-omni-30b-a3b-thinking.toml

@@ -0,0 +1,19 @@
+name = "Qwen3 Omni 30B A3B Thinking"
+description = "Qwen3 omni reasoning model for multimodal text, vision, and audio tasks"
+family = "qwen"
+release_date = "2025-09-24"
+last_updated = "2025-09-24"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 65_536
+output = 16_384
+
+[modalities]
+input = ["text", "audio", "video", "image"]
+output = ["text"]

models/alibaba/qwen3-vl-30b-a3b-instruct.toml

@@ -0,0 +1,24 @@
+# https://huggingface.co/Qwen/Qwen3-VL-30B-A3B-Instruct
+name = "Qwen3 VL 30B A3B Instruct"
+description = "Qwen vision-language instruction model for documents, visual understanding, and agent tasks"
+family = "qwen"
+release_date = "2025-10-11"
+last_updated = "2025-10-11"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 131_072
+output = 32_768
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-VL-30B-A3B-Instruct"

models/baichuan/baichuan-m2-32b.toml

@@ -0,0 +1,20 @@
+name = "Baichuan M2 32B"
+description = "Open-weight Baichuan instruction model for chat and analysis"
+family = "baichuan"
+release_date = "2025-08-13"
+last_updated = "2025-08-13"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = false
+structured_output = false
+knowledge = "2024-12"
+open_weights = true
+
+[limit]
+context = 131_072
+output = 131_072
+
+[modalities]
+input = ["text"]
+output = ["text"]

models/baidu/ernie-4.5-21b-a3b.toml

@@ -0,0 +1,20 @@
+name = "ERNIE 4.5 21B A3B"
+description = "Baidu ERNIE 4.5 open-weight mixture-of-experts instruction model"
+family = "ernie"
+release_date = "2025-06-30"
+last_updated = "2025-06-30"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+structured_output = false
+knowledge = "2025-03"
+open_weights = true
+
+[limit]
+context = 120_000
+output = 8_000
+
+[modalities]
+input = ["text"]
+output = ["text"]

providers/novita-ai/models/sao10K/l31-70b-euryale-v2.2.toml → models/baidu/ernie-4.5-300b-a47b-paddle.toml

@@ -1,20 +1,18 @@
-name = "L31 70B Euryale V2.2"
+name = "ERNIE 4.5 300B A47B"
 description = "Open-weight instruction model for adaptable chat and self-hosted production workloads"
-release_date = "2024-09-19"
-last_updated = "2024-09-19"
+family = "ernie"
+release_date = "2025-06-30"
+last_updated = "2025-06-30"
 attachment = false
 reasoning = false
 temperature = true
 tool_call = true
+structured_output = true
 open_weights = true
 
-[cost]
-input = 1.48
-output = 1.48
-
 [limit]
-context = 8_192
-output = 8_192
+context = 123_000
+output = 12_000
 
 [modalities]
 input = ["text"]

models/baidu/ernie-4.5-vl-424b-a47b.toml

@@ -0,0 +1,19 @@
+name = "ERNIE 4.5 VL 424B A47B"
+description = "Baidu ERNIE 4.5 vision-language model for visual analysis, planning, and tool use"
+family = "ernie"
+release_date = "2025-06-30"
+last_updated = "2025-06-30"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = false
+structured_output = false
+open_weights = true
+
+[limit]
+context = 123_000
+output = 16_000
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

models/deepseek/deepseek-ocr.toml

@@ -0,0 +1,25 @@
+# https://huggingface.co/deepseek-ai/DeepSeek-OCR
+name = "DeepSeek-OCR"
+description = "OCR model for extracting structured text from documents and screenshots"
+family = "deepseek"
+release_date = "2025-10-20"
+last_updated = "2025-10-20"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = false
+structured_output = true
+open_weights = true
+license = "MIT"
+
+[limit]
+context = 8_192
+output = 8_192
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/deepseek-ai/DeepSeek-OCR"

models/deepseek/deepseek-r1-0528-qwen3-8b.toml

@@ -0,0 +1,24 @@
+# https://huggingface.co/deepseek-ai/DeepSeek-R1-0528-Qwen3-8B
+name = "DeepSeek R1 0528 Qwen3 8B"
+description = "DeepSeek R1 0528 distilled into Qwen3 8B for compact math, coding, and reasoning"
+family = "deepseek-thinking"
+release_date = "2025-05-28"
+last_updated = "2025-05-28"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = false
+structured_output = false
+open_weights = true
+
+[limit]
+context = 128_000
+output = 32_000
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/deepseek-ai/DeepSeek-R1-0528-Qwen3-8B"

models/deepseek/deepseek-r1-0528.toml

@@ -0,0 +1,20 @@
+name = "DeepSeek R1 0528"
+description = "DeepSeek R1 reasoning model updated in May 2025 for math and coding"
+family = "deepseek-thinking"
+release_date = "2025-05-28"
+last_updated = "2025-05-28"
+attachment = false
+reasoning = true
+temperature = true
+knowledge = "2024-07"
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 163_840
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]

models/deepseek/deepseek-r1-distill-llama-70b.toml

@@ -0,0 +1,24 @@
+# https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B
+name = "DeepSeek R1 Distill Llama 70B"
+description = "DeepSeek R1 reasoning distilled into Llama 3.3 70B for math, coding, and analysis"
+family = "deepseek-thinking"
+release_date = "2025-01-20"
+last_updated = "2025-01-20"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = false
+structured_output = false
+open_weights = true
+
+[limit]
+context = 8_192
+output = 8_192
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B"

models/deepseek/deepseek-v3.1-terminus.toml

@@ -0,0 +1,26 @@
+# https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus
+name = "DeepSeek V3.1 Terminus"
+description = "DeepSeek V3.1 Terminus hybrid-reasoning model for coding, analysis, and agent workflows"
+family = "deepseek"
+release_date = "2025-09-22"
+last_updated = "2025-09-22"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-03-31"
+open_weights = true
+license = "MIT License"
+
+[limit]
+context = 163_840
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus"