DeepSeek V4 Pro
DeepSeek V4 Pro is DeepSeek's flagship open-source model released in April 2026, featuring a 1.6T-parameter Mixture-of-Experts architecture with 49B parameters active per token. It supports a 1M-token context window through a novel hybrid attention mechanism combining Compressed Sparse Attention and DeepSeek Sparse Attention, reducing inference FLOPs to 27% and KV cache to 10% compared to V3.2 at million-token scale. Pre-trained on 33T tokens with post-training via GRPO reinforcement learning and on-policy distillation, V4 Pro delivers frontier-level performance in coding (LiveCodeBench 93.5, Codeforces 3206), math (IMOAnswerBench 89.8), and agentic tasks (SWE-bench Verified 80.6) — competitive with GPT-5.4 and Claude Opus 4.6 at a fraction of the cost. It natively supports thinking and non-thinking modes with configurable reasoning effort, function calling, JSON output, and has been specifically optimized for mainstream agent frameworks including Claude Code, OpenClaw, and OpenCode.
Free to explore. Sign in to send messages. Sign in
LLM Playground
Chat with 133 large language models in your browser, including Claude, GPT, Gemini and DeepSeek. Tune parameters, compare answers and copy the exact API request into your code.
Why use the WaveSpeedAI LLM Playground
Leading models in one place
Switch between Claude, GPT, Gemini, DeepSeek and other leading LLMs from one picker and compare how they answer the same prompt.
Real parameters, live
Tune temperature, top_p, max_tokens and more, and turn on thinking for reasoning models. Each model shows its context window and per-token price.
Copy the exact API request
Every setting maps to the OpenAI-compatible API: copy ready-to-run cURL, Python or JavaScript and move from testing to production.
Images, video, audio and PDFs
Attach the files the selected model can read, from screenshots for vision models to PDFs, video and audio.
Sign in to send messages. Usage is billed per token at each model's listed price.
How it works
Pick a model
Search the catalog by name or provider. Each model shows its context window, capabilities and price.
Chat and tune
Send a prompt, adjust parameters or turn on thinking, and regenerate to compare answers.
Take it to code
Open API code to copy the request with your settings, then call it with your API key.
Popular models to try
Claude Opus 5.5
Anthropic · 1M context
Kimi K3.1 Pro
Moonshot · 1M context
Hy4
Tencent · 1M context
DeepSeek V4.1 Flash 1017
DeepSeek · 1M context
GPT 6.1 Sol
OpenAI · 1.1M context
Claude Sonnet 5.5
Anthropic · 1M context
Gemini 4 Carbon
Google · 1M context
Gemini 4 Argon
Google · 1M context
DeepSeek V4.1 Flash
DeepSeek · 1M context
Step 5 Preview
Stepfun · 1M context
Kimi K3
Moonshot · 1M context
GLM 5.3
ChatGLM · 1M context
Frequently asked questions
Is the LLM Playground free to use?
Opening the playground, browsing models, changing parameters and copying API code are free and need no account. Sending a message needs a WaveSpeedAI account: each request is billed per token at the model's listed input and output price, the same as an API call.
Which models can I try?
Every model in the public LLM catalog, including Claude, GPT, Gemini, DeepSeek, Qwen, Kimi, GLM and Grok. The model picker searches the whole list, and each model has its own page with pricing and details.
Can I use the same model from my code?
Yes. API code shows cURL, Python and JavaScript for the model, parameters and conversation you have set up. The endpoint is OpenAI-compatible, so the official OpenAI SDKs work once you change the base URL and API key.
Which parameters can I change?
The ones the selected model supports, such as temperature, top_p, top_k, the penalties, max_tokens and seed. Only the values you change are sent; everything else uses the model's defaults. Thinking can be turned on for models that reason before they answer.
Can I send images or files?
Text and PDF files work with every model. Vision models also accept images and videos, and models with audio input accept audio. Image links in your message are sent to vision models as images.
Is my conversation saved?
No. The conversation stays on the page while it is open; refreshing the page or starting a new chat clears it.