Disclosure: This article was written with AI assistance and reviewed by EmpirioLabs AI.
DeepSeek V4 Flash 0731 is available on EmpirioLabs. The 0731 release keeps the same 284B-total, 13B-active architecture as the preview and focuses the update on post-training for substantially stronger agent capabilities.
What changed in the 0731 release
The headline is the agent upgrade. DeepSeek reports major gains across coding, repository work, tool use, automation, and full-stack tasks, with the 0731 model outperforming the V4 Pro preview across its published agent benchmark set.
The model retains a 1M token context window and hybrid thinking, so applications can enable deeper reasoning for difficult work or disable thinking for lower-latency responses. It also supports function calling, JSON object mode, and strict JSON Schema structured output. Optional Linkup web search can add recent web context through the web_search_linkup request control.
How to call DeepSeek V4 Flash 0731
Use the OpenAI-compatible chat completions endpoint and set model to deepseek-v4-flash-0731:
curl https://api.empiriolabs.ai/v1/chat/completions \
-H 'Authorization: Bearer $EMPIRIOLABS_API_KEY' \
-H 'Content-Type: application/json' \
-d '{
"model": "deepseek-v4-flash-0731",
"enable_thinking": true,
"thinking_budget": 32768,
"messages": [
{
"role": "user",
"content": "Review this service design and return a prioritized migration plan."
}
]
}'Streaming responses include final usage data, and reasoning tokens are billed as output tokens.
Pricing
Token usage is billed at the published input and output rates. See the live model page and pricing page for current rates.
Start building
Try DeepSeek V4 Flash 0731 in the Playground, read the API documentation, or compare it with other models in the model catalog.
