DeepSeek V4 Pro 0813
DeepSeek V4 Pro 0813 is an open-weights language model from DeepSeek, released August 13, 2026. It has 1.7T parameters (mixture-of-experts, 127B active). The context window is 1M tokens. On the Epoch Capabilities Index it scores 155.9, rank 41 of 256 rated models.
DeepSeek-V4-Pro-0813 is the official release of DeepSeek-V4-Pro, superseding the preview version released 2026-04-24. Same core architecture (1.6T total / 49B active parameters) with a DSpark speculative decoding module, major agent upgrades, and low/high/max reasoning settings. Open weights (MIT) on Hugging Face.
Check it against your own PC →
Can you run DeepSeek V4 Pro 0813 locally?
Even with 32 GB of VRAM and 32 GB of RAM it is too large to run at a usable speed on a single consumer GPU. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Too large | – | – | – | – |
| 8 GB VRAM | Too large | – | – | – | – |
| 12 GB VRAM | Too large | – | – | – | – |
| 16 GB VRAM | Too large | – | – | – | – |
| 24 GB VRAM | Too large | – | – | – | – |
| 32 GB VRAM | Too large | – | – | – | – |
Ready-made quantizations on Hugging Face: unsloth/DeepSeek-V4-Pro-0813-GGUF · qtum/DeepSeek-V4-Pro-0813-GGUF · 6block/DeepSeek-V4-Pro-0813-GGUF · pokey2025/dspark-DeepSeek-V4-Pro-0813.gguf · ryant0000/DeepSeek-V4-Pro-0813-gguf · DevQuasar/deepseek-ai.DeepSeek-V4-Pro-0813-GGUF
Strengths by category
ModelTape's 0–100 fit on Epoch AI benchmark results.
| Category | Score | Based on |
|---|---|---|
| Reasoning | 26.6 | Chess Puzzles, Mystery Game Puzzles |
| Math | 52.1 | ProofBench, OTIS Mock AIME 2024-2025, FrontierMath-Tiers-1-3-v2-Private, FrontierMath-Tier-4-v2-Private |
| Knowledge | 77.3 | GPQA diamond, SimpleQA Verified |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| OTIS Mock AIME | 98.6% |
| GPQA Diamond | 92.6% |
| FrontierMath (Tiers 1–3) | 68.0% |
| SimpleQA Verified | 52.9% |
| ProofBench | 50.0% |
| Chess Puzzles | 43.2% |
| Mystery Game Puzzles | 37.2% |
| MirrorCode | 33.3% |
| FrontierMath-Tier-4-v2-Private | 26.8% |
Links
Announcement · Hugging Face: deepseek-ai/DeepSeek-V4-Pro-0813 · Epoch AI benchmarks
Models with a similar score
Claude Opus 4.8 · DeepSeek V4.1 Flash 1017 · Gemini 3.8 Flash · Kimi K3 · Gemini 3.7 Flash · Muse Spark 1.3 · GPT-5.4 · Grok 4.6
More from DeepSeek
DeepSeek V4.1 Flash 1017 · DeepSeek V4 Flash Vision Exp · DeepSeek V4.1 Flash · DeepSeek V4 Flash 0731 · DeepSeek V4 Pro DSpark · DeepSeek V4 Flash DSpark · DeepSeek-V4-Pro · DeepSeek-V4-Flash