DeepSeekopen-weight
Released: 2026-08-16DeepSeek V4-Pro
DeepSeek 1.6 trillion parameter flagship open-weight MoE model with DeepSeek Sparse Attention (DSA) and 1M token context.
Intelligence Index
92.2/ 100
Calibrated multi-domain compositeThroughput Speed
65tok/sec
Output streaming throughputTime To First Token
390ms
Initial chunk server latencyLiveBench Contam-Free
77.9%
September 2026 suite scoreLiveBench Multi-Domain Evaluation Breakdown
Monthly refreshed contamination-free test setsLogical Reasoning87.1%
Coding & Repo Repair81.0%
Mathematics (AIME)91.4%
Data Analysis & Tables75.2%
Language & Comprehension70.1%
Instruction Following64.8%
Sourced Benchmark Evaluations (3)
Verified performance across authoritative benchmarks with provenance tracking.Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run deepseek-v4-pro
# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4-pro",
"messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
}'Evaluation & Sourcing Notes
Scores listed for DeepSeek V4-Pro represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.