DeepSeek V4 Pro 0813
by DeepSeek Expected Sep 20, 2026
Coming soonopen-weightsreasoninglarge
thairouter/deepseek-v4-proOverview
DeepSeek's flagship open-weights reasoning model. Top-tier on coding, math and agentic benchmarks at a fraction of proprietary prices.
Context
131.1K
tokens
Max output
8.2K
tokens per reply
Input
TBA
per 1M tokens
Output
TBA
per 1M tokens
Expected Sep 20, 2026. Specs are provisional until the model is actually served; pricing is announced at launch.
Benchmarks
Reference scores published by the model's authors, not measured on ThaiRouter. Live numbers appear once the model is serving.
Intelligence Index
36
Artificial Analysis v4.3
API
OpenAI-compatible. Point your client at https://api.thairouter.ai/v1 and send this id.
Thinking effort
Default high. Can be switched off. Thinking tokens are billed as output.
curl https://api.thairouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THAIROUTER_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "thairouter/deepseek-v4-pro",
"messages": [{"role": "user", "content": "สวัสดี"}],
"reasoning_effort": "high"
}'Not serving yet. Requests with this id return 404 until launch.