@apteva/integrations 0.3.62 → 0.15.4

package/src/apps/venice-ai.json

@@ -0,0 +1,685 @@
+{
+  "slug": "venice-ai",
+  "name": "Venice AI",
+  "description": "Privacy-first, OpenAI-compatible API for chat, image generation (Grok Imagine, Nano Banana, Venice SD 3.5, gpt-image-2), text-to-speech (multiple TTS engines + voice cloning handles), and image upscaling. Bearer-token auth; no data retention on Venice's side.",
+  "logo": "https://www.google.com/s2/favicons?domain=venice.ai&sz=128",
+  "categories": [
+    "ai",
+    "venice",
+    "chat",
+    "image-generation",
+    "tts",
+    "upscale",
+    "uncensored"
+  ],
+  "base_url": "https://api.venice.ai/api/v1",
+  "auth": {
+    "types": [
+      "bearer"
+    ],
+    "headers": {
+      "Authorization": "Bearer {{token}}"
+    },
+    "credential_fields": [
+      {
+        "name": "token",
+        "label": "API Key",
+        "description": "Your Venice AI API Key — create one at https://venice.ai/settings/api"
+      }
+    ]
+  },
+  "tools": [
+    {
+      "name": "generate_image",
+      "description": "Generate an image via Venice's POST /image/generate. Supports grok-imagine-image, gpt-image-2, venice-sd35, qwen-image-2, nano-banana-pro, and others (call list_models?type=image for the live set). Returns JSON {id, images:[<b64>], timing} by default (return_binary=false). Most Venice-specific knobs (style_preset, hide_watermark, safe_mode, variants) are exposed.",
+      "method": "POST",
+      "path": "/image/generate",
+      "timeout_ms": 600000,
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "model": {
+            "type": "string",
+            "description": "Image model id — e.g. grok-imagine-image (default), gpt-image-2, venice-sd35, qwen-image-2, nano-banana-pro. Use list_models?type=image for the authoritative list.",
+            "default": "grok-imagine-image"
+          },
+          "prompt": {
+            "type": "string",
+            "description": "Image description (1–7500 chars, model-specific limit)."
+          },
+          "negative_prompt": {
+            "type": "string",
+            "description": "Things to steer the image away from (up to 7500 chars)."
+          },
+          "width": {
+            "type": "integer",
+            "description": "Image width in pixels (1–1280). Pixel-sized models (venice-sd35) honour this; aspect-ratio models (nano-banana, qwen-image-2) ignore it — use aspect_ratio + resolution instead.",
+            "minimum": 1,
+            "maximum": 1280,
+            "default": 1024
+          },
+          "height": {
+            "type": "integer",
+            "description": "Image height in pixels (1–1280).",
+            "minimum": 1,
+            "maximum": 1280,
+            "default": 1024
+          },
+          "aspect_ratio": {
+            "type": "string",
+            "description": "Aspect ratio (aspect-ratio models only — nano-banana, qwen-image-2). e.g. 1:1, 16:9, 3:2.",
+            "enum": [
+              "1:1",
+              "3:2",
+              "2:3",
+              "16:9",
+              "9:16",
+              "4:3",
+              "3:4",
+              "21:9"
+            ]
+          },
+          "resolution": {
+            "type": "string",
+            "description": "Resolution tier (resolution-tier models only — gpt-image-2, nano-banana-pro).",
+            "enum": [
+              "1K",
+              "2K",
+              "4K"
+            ]
+          },
+          "format": {
+            "type": "string",
+            "description": "Output image format.",
+            "enum": [
+              "jpeg",
+              "png",
+              "webp"
+            ],
+            "default": "webp"
+          },
+          "cfg_scale": {
+            "type": "number",
+            "description": "Adherence to prompt (0–20, model-specific defaults). Higher = more literal, lower = more creative.",
+            "minimum": 0,
+            "maximum": 20
+          },
+          "steps": {
+            "type": "integer",
+            "description": "Diffusion steps (model-specific default; some models ignore this).",
+            "default": 8
+          },
+          "seed": {
+            "type": "integer",
+            "description": "Random seed for reproducibility (-999999999 to 999999999). 0 = random.",
+            "minimum": -999999999,
+            "maximum": 999999999,
+            "default": 0
+          },
+          "style_preset": {
+            "type": "string",
+            "description": "Style preset — e.g. '3D Model', 'Photographic', 'Anime', 'Digital Art'. Model-specific support."
+          },
+          "safe_mode": {
+            "type": "boolean",
+            "description": "Blur adult-classified content. Default true.",
+            "default": true
+          },
+          "hide_watermark": {
+            "type": "boolean",
+            "description": "Suppress the Venice watermark. Venice may override for certain content.",
+            "default": false
+          },
+          "return_binary": {
+            "type": "boolean",
+            "description": "Return raw image bytes instead of JSON+base64. Default false (returns JSON with base64 image strings).",
+            "default": false
+          },
+          "variants": {
+            "type": "integer",
+            "description": "Number of image variants to generate (1–4). Only honoured when return_binary=false.",
+            "minimum": 1,
+            "maximum": 4,
+            "default": 1
+          },
+          "lora_strength": {
+            "type": "integer",
+            "description": "LoRA strength when the chosen model supports adapters (0–100).",
+            "minimum": 0,
+            "maximum": 100
+          },
+          "embed_exif_metadata": {
+            "type": "boolean",
+            "description": "Embed generation metadata in the image EXIF block.",
+            "default": false
+          }
+        },
+        "required": [
+          "model",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "chat_completion",
+      "description": "OpenAI-compatible chat completion via POST /chat/completions. Venice extras (web search, character_slug, thinking control) ride on `venice_parameters`.",
+      "method": "POST",
+      "path": "/chat/completions",
+      "timeout_ms": 120000,
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "model": {
+            "type": "string",
+            "description": "Text model id — e.g. venice-uncensored, llama-3.3-70b, qwen-2.5-coder-32b, deepseek-r1-671b. Use list_models?type=text for the live set."
+          },
+          "messages": {
+            "type": "array",
+            "description": "OpenAI-shaped messages array.",
+            "items": {
+              "type": "object",
+              "properties": {
+                "role": {
+                  "type": "string",
+                  "enum": [
+                    "system",
+                    "user",
+                    "assistant",
+                    "tool"
+                  ]
+                },
+                "content": {
+                  "type": "string"
+                }
+              },
+              "required": [
+                "role",
+                "content"
+              ]
+            }
+          },
+          "max_tokens": {
+            "type": "integer",
+            "description": "Cap on response tokens.",
+            "minimum": 1
+          },
+          "temperature": {
+            "type": "number",
+            "description": "Sampling temperature (0–2).",
+            "minimum": 0,
+            "maximum": 2
+          },
+          "top_p": {
+            "type": "number",
+            "description": "Nucleus sampling (0–1).",
+            "minimum": 0,
+            "maximum": 1
+          },
+          "stream": {
+            "type": "boolean",
+            "description": "Stream tokens via SSE.",
+            "default": false
+          },
+          "venice_parameters": {
+            "type": "object",
+            "description": "Venice-specific knobs.",
+            "properties": {
+              "include_venice_system_prompt": {
+                "type": "boolean",
+                "default": true
+              },
+              "character_slug": {
+                "type": "string",
+                "description": "Use a Venice Character persona."
+              },
+              "enable_web_search": {
+                "type": "string",
+                "enum": [
+                  "off",
+                  "on",
+                  "auto"
+                ],
+                "default": "off"
+              },
+              "enable_web_scraping": {
+                "type": "boolean",
+                "default": false
+              },
+              "enable_web_citations": {
+                "type": "boolean",
+                "default": false
+              },
+              "strip_thinking_response": {
+                "type": "boolean",
+                "default": false
+              },
+              "disable_thinking": {
+                "type": "boolean",
+                "default": false
+              }
+            }
+          }
+        },
+        "required": [
+          "model",
+          "messages"
+        ]
+      }
+    },
+    {
+      "name": "text_to_speech",
+      "description": "Generate speech audio via POST /audio/speech. Returns binary audio (Content-Type matches `response_format`). Supports voice cloning handles (vv_<slug>) alongside the curated voice catalog.",
+      "method": "POST",
+      "path": "/audio/speech",
+      "timeout_ms": 120000,
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "model": {
+            "type": "string",
+            "description": "TTS model id.",
+            "enum": [
+              "tts-kokoro",
+              "tts-qwen3-0-6b",
+              "tts-qwen3-1-7b",
+              "tts-xai-v1",
+              "tts-chatterbox-hd",
+              "tts-elevenlabs-turbo-v2-5",
+              "tts-minimax-speech-02-hd",
+              "tts-gemini-3-1-flash"
+            ],
+            "default": "tts-kokoro"
+          },
+          "input": {
+            "type": "string",
+            "description": "Text to speak (1–4096 chars)."
+          },
+          "voice": {
+            "type": "string",
+            "description": "Voice id — curated catalog (af_sky, af_bella, af_nova, am_adam, am_echo, bm_daniel, eve, Vivian, Rachel, Aria, …) OR a cloned-voice handle matching vv_<slug>.",
+            "default": "af_sky"
+          },
+          "response_format": {
+            "type": "string",
+            "description": "Audio container.",
+            "enum": [
+              "mp3",
+              "opus",
+              "aac",
+              "flac",
+              "wav",
+              "pcm"
+            ],
+            "default": "mp3"
+          },
+          "speed": {
+            "type": "number",
+            "description": "Playback speed multiplier.",
+            "minimum": 0.25,
+            "maximum": 4,
+            "default": 1
+          }
+        },
+        "required": [
+          "input"
+        ]
+      }
+    },
+    {
+      "name": "edit_image",
+      "description": "Prompt-driven image edit via POST /image/edit. Short, instructional prompts work best ('remove the tree', 'change the sky to sunrise'). Returns binary image bytes in the requested output_format. Models include firered-image-edit (default), qwen-edit, grok-imagine-edit, flux-2-max-edit, gpt-image-2-edit, and ~9 more (call list_models?type=image for the live set; edit-capable models surface there).",
+      "method": "POST",
+      "path": "/image/edit",
+      "timeout_ms": 180000,
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "image": {
+            "type": "string",
+            "description": "Source image — file upload, base64 string, or http(s) URL. Dimensions: 65,536–33,177,600 pixels."
+          },
+          "prompt": {
+            "type": "string",
+            "description": "Edit instruction (≤32,768 chars). Short imperatives perform best."
+          },
+          "model": {
+            "type": "string",
+            "description": "Edit model id. Default firered-image-edit.",
+            "default": "firered-image-edit"
+          },
+          "aspect_ratio": {
+            "type": "string",
+            "description": "Output aspect ratio. 'auto' keeps the source's ratio.",
+            "enum": [
+              "auto",
+              "1:1",
+              "3:2",
+              "2:3",
+              "16:9",
+              "9:16",
+              "21:9",
+              "3:4",
+              "4:5"
+            ],
+            "default": "auto"
+          },
+          "resolution": {
+            "type": "string",
+            "description": "Output resolution tier.",
+            "enum": [
+              "1K",
+              "2K",
+              "4K"
+            ],
+            "default": "1K"
+          },
+          "output_format": {
+            "type": "string",
+            "description": "Output image format.",
+            "enum": [
+              "jpeg",
+              "png",
+              "webp"
+            ],
+            "default": "png"
+          },
+          "safe_mode": {
+            "type": "boolean",
+            "description": "Blur adult-classified content.",
+            "default": true
+          }
+        },
+        "required": [
+          "image",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "multi_edit_image",
+      "description": "Prompt-driven multi-image edit via POST /image/multi-edit. Accepts 1–3 reference images as base64 strings or http(s) URLs and returns binary image bytes in the requested output_format. Use this when a caller supplies multiple persona, style, item, or scene references.",
+      "method": "POST",
+      "path": "/image/multi-edit",
+      "timeout_ms": 180000,
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "images": {
+            "type": "array",
+            "description": "Source images — base64 strings or http(s) URLs. Venice accepts 1–3 images.",
+            "minItems": 1,
+            "maxItems": 3,
+            "items": {
+              "type": "string"
+            }
+          },
+          "prompt": {
+            "type": "string",
+            "description": "Edit instruction (≤32,768 chars)."
+          },
+          "modelId": {
+            "type": "string",
+            "description": "Edit model id. Default firered-image-edit.",
+            "default": "firered-image-edit"
+          },
+          "aspect_ratio": {
+            "type": "string",
+            "description": "Output aspect ratio.",
+            "enum": [
+              "auto",
+              "1:1",
+              "3:2",
+              "2:3",
+              "16:9",
+              "9:16",
+              "21:9",
+              "3:4",
+              "4:5"
+            ],
+            "default": "auto"
+          },
+          "quality": {
+            "type": "string",
+            "description": "Output quality for models that expose quality tiers.",
+            "enum": [
+              "low",
+              "medium",
+              "high",
+              "auto"
+            ],
+            "default": "auto"
+          },
+          "resolution": {
+            "type": "string",
+            "description": "Output resolution tier.",
+            "enum": [
+              "1K",
+              "2K",
+              "4K"
+            ],
+            "default": "1K"
+          },
+          "output_format": {
+            "type": "string",
+            "description": "Output image format.",
+            "enum": [
+              "jpeg",
+              "png",
+              "webp"
+            ],
+            "default": "png"
+          },
+          "safe_mode": {
+            "type": "boolean",
+            "description": "Blur adult-classified content.",
+            "default": true
+          }
+        },
+        "required": [
+          "images",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "upscale_image",
+      "description": "Upscale or enhance an image via POST /image/upscale. Inputs accept base64 or remote URL bytes (≤25 MB, ≥65536 px, ≤16777216 px after scaling). Returns binary PNG.",
+      "method": "POST",
+      "path": "/image/upscale",
+      "timeout_ms": 180000,
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "image": {
+            "type": "string",
+            "description": "Source image — base64 string or fetchable URL."
+          },
+          "scale": {
+            "type": "number",
+            "description": "Output multiplier (1–4). scale=1 requires enhance=true.",
+            "minimum": 1,
+            "maximum": 4,
+            "default": 2
+          },
+          "enhance": {
+            "type": "boolean",
+            "description": "Apply enhancement pass. Required when scale=1.",
+            "default": false
+          },
+          "enhanceCreativity": {
+            "type": "number",
+            "description": "How much the enhancer may invent (0–1).",
+            "minimum": 0,
+            "maximum": 1,
+            "default": 0.5
+          },
+          "enhancePrompt": {
+            "type": "string",
+            "description": "Optional style hint for the enhancer (≤1500 chars)."
+          },
+          "replication": {
+            "type": "number",
+            "description": "How faithfully to preserve the source (0–1).",
+            "minimum": 0,
+            "maximum": 1,
+            "default": 0.35
+          }
+        },
+        "required": [
+          "image"
+        ]
+      }
+    },
+    {
+      "name": "queue_video",
+      "description": "Queue a video generation job via POST /video/queue. Async — returns {queue_id, model} immediately; poll retrieve_video to get the bytes. Supports text-to-video, image-to-video (image_url), and a rich reference-asset surface. Models: kling-2, veo-3-1, ltx-video-2, runway-gen-4-5, hailuo-02, seedance-2, … (call list_models?type=video for the live set).",
+      "method": "POST",
+      "path": "/video/queue",
+      "timeout_ms": 60000,
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "model": {
+            "type": "string",
+            "description": "Video model id."
+          },
+          "prompt": {
+            "type": "string",
+            "description": "Description (1–10,000 chars, model-specific cap)."
+          },
+          "duration": {
+            "type": "string",
+            "description": "Video length: '2s'–'30s', '1 gen', or 'Auto'. Per-model support varies."
+          },
+          "negative_prompt": {
+            "type": "string"
+          },
+          "aspect_ratio": {
+            "type": "string",
+            "enum": ["1:1", "2:3", "3:2", "3:4", "4:3", "9:16", "16:9", "21:9"]
+          },
+          "resolution": {
+            "type": "string",
+            "description": "Output resolution — 256p..4k, 2x, 4x, true_1080p (model-dependent)."
+          },
+          "upscale_factor": {
+            "type": "integer",
+            "description": "1, 2, or 4 — upscale-capable models only.",
+            "enum": [1, 2, 4]
+          },
+          "audio": {
+            "type": "boolean",
+            "description": "Generate audio track (audio-capable models). Default true.",
+            "default": true
+          },
+          "image_url": {
+            "type": "string",
+            "description": "Image-to-video source — HTTPS URL or data: URL."
+          },
+          "end_image_url": {
+            "type": "string",
+            "description": "End-frame target — same encoding as image_url."
+          },
+          "audio_url": {
+            "type": "string",
+            "description": "Reference audio for lip-sync / score (WAV/MP3, ≤30s, ≤15 MB)."
+          },
+          "video_url": {
+            "type": "string",
+            "description": "Video-to-video / upscale source (MP4/MOV/WebM)."
+          },
+          "reference_image_urls": {
+            "type": "array",
+            "items": { "type": "string" },
+            "description": "Up to 9 reference images (character / style consistency)."
+          },
+          "reference_video_urls": {
+            "type": "array",
+            "items": { "type": "string" },
+            "description": "Up to 3 reference videos (motion / style; 2–15s each, ≤50 MB)."
+          },
+          "reference_audio_urls": {
+            "type": "array",
+            "items": { "type": "string" },
+            "description": "Up to 3 reference audio clips for vocal timbre (2–15s each)."
+          }
+        },
+        "required": ["model", "prompt", "duration"]
+      }
+    },
+    {
+      "name": "retrieve_video",
+      "description": "Poll a queued video job via POST /video/retrieve. While processing returns JSON {status:'PROCESSING', average_execution_time, execution_duration}. When complete returns binary mp4 (wrapped as {_binary, base64, mimeType} by the executor).",
+      "method": "POST",
+      "path": "/video/retrieve",
+      "timeout_ms": 120000,
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "queue_id": { "type": "string", "description": "From queue_video." },
+          "model": { "type": "string", "description": "Same model id as the queue call." },
+          "delete_media_on_completion": { "type": "boolean", "default": false }
+        },
+        "required": ["queue_id", "model"]
+      }
+    },
+    {
+      "name": "quote_video",
+      "description": "Get a USD price quote for a video gen BEFORE queueing via POST /video/quote. Returns {quote: <number>} — useful for cost-aware UIs (panel can show 'this will cost $X' before submit).",
+      "method": "POST",
+      "path": "/video/quote",
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "model": { "type": "string" },
+          "duration": { "type": "string" },
+          "aspect_ratio": {
+            "type": "string",
+            "enum": ["1:1", "2:3", "3:2", "3:4", "4:3", "9:16", "16:9", "21:9"]
+          },
+          "resolution": { "type": "string" },
+          "upscale_factor": { "type": "integer", "enum": [1, 2, 4] },
+          "audio": { "type": "boolean" },
+          "video_url": { "type": "string", "description": "Base64 video for upscaling; auto-detects duration." },
+          "reference_video_total_duration": {
+            "type": "number",
+            "description": "Aggregate reference-video seconds for R2V models (max 45)."
+          }
+        },
+        "required": ["model", "duration"]
+      }
+    },
+    {
+      "name": "list_models",
+      "description": "List available models via GET /models. Filter by `type` — image | text | tts | video | music | upscale | inpaint | embedding | asr | code | all.",
+      "method": "GET",
+      "path": "/models",
+      "input_schema": {
+        "type": "object",
+        "properties": {
+          "type": {
+            "type": "string",
+            "description": "Model family filter. 'all' returns every model.",
+            "enum": [
+              "all",
+              "image",
+              "text",
+              "tts",
+              "video",
+              "music",
+              "upscale",
+              "inpaint",
+              "embedding",
+              "asr",
+              "code"
+            ],
+            "default": "all"
+          }
+        }
+      }
+    }
+  ],
+  "health_check": {
+    "tool": "list_models"
+  }
+}