Bytedance Seed V2 Mini

Large Language Models

Seed 2.0 Mini is a high-performance multimodal model optimized for low latency and high concurrency. It supports text, image, and video input with 256K context and configurable thinking/reasoning modes.

Inference$ Commercial useFrom 1 credit per run

About Bytedance Seed V2 Mini

Pricing

A run with the default settings costs 1 credits. The exact cost updates in the playground as you change settings, and you're only charged when you press Run. Failed runs are refunded automatically.

Inputs

prompttextarearequired

The text prompt or question for the model.

image_urlsfile-list

URLs of images for visual understanding. Supported formats: JPEG, PNG, WebP. A maximum of 6 images is supported. Any additional images will be ignored.

video_urlsfile-list

URLs of videos for video understanding. Supported formats: MP4, MOV. Audio comprehension is not supported. A maximum of 3 videos is supported. Any additional videos will be ignored.

system_prompttextarea

Optional system prompt to guide the model's behavior.

messagesjson

Optional prior conversation history for multi-turn conversations. Pass back the `messages` field from a previous response to provide context. The current `prompt`, `image_urls`, `video_urls`, and `system_prompt` are always appended as the latest user turn.

thinkingselectdefault: "enabled"

Controls the model's chain-of-thought reasoning. `enabled` always includes reasoning, `disabled` never includes reasoning, `auto` lets the model decide based on the query.

Options: enabled, disabled, auto

reasoning_effortselect

Controls the depth of reasoning before the model responds. Only applicable when `thinking` is `enabled` or `auto`. `minimal` for immediate response, `low` for faster response with light reasoning, `medium` for balanced speed and depth, `high` for deep analysis of complex issues.

Options: minimal, low, medium, high

max_completion_tokensintegerdefault: 4096

Controls the maximum length of the model's output, including both the model's response and its chain-of-thought content, measured in tokens.

temperaturenumberdefault: 1

Controls randomness in the response. Lower values make output more focused and deterministic, higher values make it more creative.

top_pnumberdefault: 0.7

Nucleus sampling parameter. The model considers tokens with top_p cumulative probability mass. Lower values narrow the token selection.

enable_safety_checkerbooleandefault: true

Authorized callers may set false to request relaxed screening.

Related large language models models

OpenRouter Chat Completions [OpenAI Compatible]

OpenAI-compatible chat completions API. Drop-in replacement for the OpenAI API — use any OpenAI SDK or client to access Claude, Gemini, Grok, DeepSeek, Llama, Qwen, Mistral, and all OpenAI models (GPT-5, GPT-4o, o3) through fal. Powered by OpenRouter.

large-language-models

OpenRouter

Run any LLM with fal. Access Claude (Anthropic), ChatGPT / GPT-5 / GPT-4o (OpenAI), Gemini (Google), Grok (xAI), DeepSeek, Llama (Meta), Qwen (Alibaba), Mistral, and 200+ more models through a single API. Supports reasoning, structured output, and streaming. Powered by OpenRouter.

large-language-models

OpenRouter Responses [OpenAI Compatible]

The OpenRouter Responses API with fal, powered by OpenRouter, provides unified access to a wide range of large language models - including GPT, Claude, Gemini, and many others through a single API interface.

large-language-models

OpenRouter Embeddings [OpenAI Compatible]

Generate text embeddings using OpenAI-compatible API. Access embedding models like text-embedding-3-small, text-embedding-3-large (OpenAI), and other embedding models available through OpenRouter. Drop-in replacement for the OpenAI embeddings API. Powered by OpenRouter.

large-language-models

OpenRouter [Enterprise]

Run any LLM (Large Language Model) with fal, powered by OpenRouter.

large-language-models

✦ new

NVIDIA Nemotron 3.5 Super Text

Chat and reason with NVIDIA Nemotron 3.5 Super, a hybrid Mamba-Transformer mixture-of-experts model with optional step-by-step thinking.

reasoningchat

large-language-models

Nemotron 3 Nano Omni

Open, efficient reasoning model from NVIDIA. 30B A3B hybrid Transformer-Mamba MoE, built for enterprise agentic workflows.

nemotronnvidia

large-language-models