#16DeepSeek· released 2026-04-24

DeepSeek: DeepSeek V4 Pro 0423.

DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with 1.6T total parameters and 49B activated parameters, supporting a 1M-token context window. It is designed for advanced reasoning, coding,...

View on OpenRouter ↗
Context
1.0M
Input $/1M
$0.78
Output $/1M
$1.56
Modality
text
Arena Elo
1562
LMArena — human blind votes
AA Index
64.0
Artificial Analysis composite
LiveBench
81.7
Contamination-resistant
GPQA
84.9%
Grad-level science reasoning
MMLU-Pro
84.5%
Broad knowledge
SWE-bench
70.2%
Real GitHub bugs (Verified)
Terminal-Bench
57.1%
Agentic shell tasks
ARC-AGI
17.2%
Abstract reasoning
Aider Polyglot
77.4%
Multi-language code editing
Speed
82 tok/s
Median output speed

Profile

DeepSeek: DeepSeek V4 Pro 0423 is DeepSeek's entry at position #16 on the current aggregated leaderboard. Its strongest signals are graduate-level science reasoning, real-world software engineering, unattended agentic work, and very long context. Every figure below is aggregated from the public sources listed on the homepage and refreshed on each pipeline run.

The score spread is comparatively complete for this model, but the standing advice still applies: treat gaps of a few points as ties and validate on your own workload before committing.

How this model's position is computed is documented in the methodology; what each benchmark actually measures is in the benchmark guide. See a figure that disagrees with its source? Report it — accepted corrections apply on the next refresh.

Nearby on the leaderboard