whatmodel.sh / DeepSeek V4 Pro (0813)

DeepSeek V4 Pro (0813)

+ contribute

DeepSeek V4 Pro (0813) is an open-weights language model from DeepSeek, released August 2026. Its published size is 1600B parameters (49B active per token) and a 1M-token context window. Licence: mit.

76.4±0.7#31 of 719 LLMsfrom 31 benchmark results across 8 sources

$ how it compares

The LLMs ranked either side of DeepSeek V4 Pro (0813), out of 828 with a score, then more from DeepSeek.

#	model	score	where	RAM	released
29	GPT-5.6 Luna OpenAI	77±1	api $0.45/M		2026-07
30	Grok 4.6 SpaceXAI (xAI)	77±1	api $3.00/M		2026-08
31	DeepSeek V4 Pro (0813) DeepSeek	76±1	api $1.63/M +local	~1010GB	2026-08
32	Gemini 3.7 Flash Google	76±1	api $1.50/M		2026-08
33	GLM-5.3 Z.ai	76±1	api $2.15/M +local	~239GB	2026-08
	more from DeepSeek
21	DeepSeek V4.1 Flash DeepSeek	78±1	api $0.26/M +local	~537GB	2026-09
35	DeepSeek V3.2 Speciale DeepSeek	76±3	comparison only
36	DeepSeek V4 Flash Vision (Exp) DeepSeek	76±2	local	~104GB	2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #31 of 300 · 24% of its score · 10 tests

Beats GPT-5.4, GPT-6 Luna, GPT-5.6 Luna here, though they rank higher overall. Trails GLM-5.3-Flash, Grok 4.7, Qwen3.8 Max (0902) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Arena Agent (October 2026 aggregate IPS)	0.0119 IPS ✎	▁▁▁▄▂▃▇█▂▂▁▃ beats 23/49	avg -0.312 IPS	▇▇▇ 7.6%	sharp
GameLogicBench solve rate (72 Godot game-logic tasks, Claude Code scaffold)	41.7% ✎	█▄·█·▄·██▄·▄ beats 8/11	very hard · avg 4.7%	▇ 3.1%	at its level
Terminal-Bench 4.0 lab	12.4% ✎	▁·▁··▁▇▇▅██▃ beats 4/30	very hard · avg 0.3%	▇ 3.1%	sharp · small run
Terminal-Bench 2.1 lab	87.9% ✎	▁·▁·▁▁▄▃▄█▃▇ beats 52/56	hard · avg 26%	▇ 2.7%	small run
Terminal-Bench Hard	51.4% ✎	▂▅▆▂▂▆▂▂█▅▅▃ beats 27/27	very hard · avg 4.3%	▇ 2.5%	sharp · at its level · small run
Humanity's Last Exam with tools lab	60.0% ✎	▂▂▂▁▄▂▃▆█▇▃▆ beats 47/54	hard · avg 17%	▇ 1.3%	at its level · small run
DeepSWE v1.1 (resolved) lab	62.7% ✎	▂····▁▁▁▆██▆ beats 9/28	hard · avg 21%	▇ 1.1%	at its level · small run
Agents' Last Exam (pass@1, official scaffold) lab	25.7% ✎	█··········▄ beats 1/2	very hard · avg 6.2%	▇ 1.0%	at its level · small run

2 more tests

test	result	vs the others in its run	difficulty	counts	why
Toolathlon-Verified lab	74.1% ✎	▂·▂▂▅▃···▂██ beats 10/18	hard · avg 33%	▇ 0.7%	blunt · at its level · small run
Terminal-Bench 3.0 lab	11.8% ✎	█··█·······█ beats 1/2	very hard · avg 2.1%	▇ 0.6%	noisy · small run

coding▁▁▁▃▆██▇▆▄▃▁ #28 of 458 · 13% of its score · 4 tests

Beats GPT-5.5, GPT-5.6 Terra, Claude Opus 4.8 here, though they rank higher overall. Trails GLM-5.2, GLM-5.3-Flash, Grok 4.7 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Arena WebDev (Code Arena)	1582.6 ✎	▁▂▄▃▅█▄▇▆▄▁▁ beats 91/119	avg 981	▇▇▇ 6.7%	sharp
GameLogicBench solve rate (72 Godot game-logic tasks, Claude Code scaffold)	41.7% ✎	█▄·█·▄·██▄·▄ beats 8/11	very hard · avg 4.7%	▇ 3.1%	at its level
Luau Leetcode fully solved (Torpedo Software)	82.2% ✎	▁▁▁▁·▁▂▃▅▃█▇ beats 37/49	hard · avg 18%	▇ 2.4%	typical
DeepSWE v1.1 (resolved) lab	62.7% ✎	▂····▁▁▁▆██▆ beats 9/28	hard · avg 21%	▇ 1.1%	at its level · small run

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #38 of 393 * · 3.8% of its score · 1 test

Beats GPT-6 Sol, MiMo-V2.6-Flash (309B-A15B), GPT-6 Luna here, though they rank higher overall. Trails MiMo-V2.5-Pro (1T-A42B), Grok 4.20 (0309 reasoning), Claude Opus 4.5 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
LMArena (text)	1464.3 ✎	▁▁▁▂▃▃▃▆▆▇█▃ beats 333/379	avg 1268	▇▇ 3.8%	blunt

math▁▂▃▅▆▅▄█▄▁▁▁ #33 of 398 · 35% of its score · 4 tests

Trails GPT-5.2, Qwen3.7 Max, Gemini 3.1 Pro (Preview) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
FrontierMath tier 4 v2	26.8% ✎	▁▂▁▂█▅▂▂▁▁▁▁ beats 32/63	very hard · avg 0.1%	▇▇▇▇▇▇▇ 17%	sharp · at its level
FrontierMath tiers 1–3 v2	64.6% ✎	▂··▁▁▃▄▄█▅▃▃ beats 50/82	very hard · avg 1.2%	▇▇▇▇▇▇ 16%	sharp · at its level
MathArena Apex (pass@1) lab	65.3% ✎	▄····▄···▄█▄ beats 3/5	very hard · avg 15%	▇ 1.7%	at its level · small run
OTIS mock AIME 2024–2025	98.6% ✎	▃▃▄▃▃▃▇▇█▃▃▅ beats 176/193	very hard · avg 8.5%	▇ 0.6%	sharp · maxed out · noisy

web search██·········█ #2 of 3 * · 1.3% of its score · 1 test

test	result	vs the others in its run	difficulty	counts	why
Humanity's Last Exam with tools lab	60.0% ✎	▂▂▂▁▄▂▃▆█▇▃▆ beats 47/54	hard · avg 17%	▇ 1.3%	at its level · small run

other tests14 not in a task · 28% of its score

test	result	vs the others in its run	difficulty	counts	why
GPQA Diamond +1	91.7% ✎	▁▁▁▁▂▄▆▄▅▇█▅ beats 186/208	hard · avg 39%	▇▇▇ 6.4%	sharp · small run · several runs
Mystery Game Puzzles (Epoch)	43.0% ✎	█··▂▃▇▄█▃▃▁▁ beats 63/74	very hard · avg 9.4%	▇▇ 5.6%	sharp · noisy
Chess Puzzles (Epoch)	47.0% ✎	█▁▁▁▁▂▁▄▂▃▁▁ beats 125/135	very hard · avg 5.9%	▇ 3.4%	sharp · at its level · noisy
Humanity's Last Exam lab	42.7% ✎	▁▁▁▃▆▆▇██▆▁▃ beats 93/104	very hard · avg 5.3%	▇ 2.1%	at its level · small run
NL2Repo lab	61.5% ✎	▂▂·▂▂▄█▆▆▂▄▆ beats 20/21	very hard · avg 13%	▇ 1.8%	at its level · small run
SimpleQA Verified	52.9% ✎	▁▁▃▂▃▃▆▆█▃▂▅ beats 58/78	very hard · avg 8.0%	▇ 1.5%	at its level
SEC-Bench Pro (pass@1) lab	56.4% ✎	▄··▄···▄·▄·█ beats 3/5	very hard · avg 7.2%	▇ 1.4%	at its level · small run
Icelandic LLM Leaderboard average (Miðeind, Sept 2026 suite)	51.2% ✎	▁▁▃▂▂▄▁▃▄▄▄█ beats 20/38	very hard · avg 14%	▇ 1.3%	at its level

6 more tests

test	result	vs the others in its run	difficulty	counts	why
CyberGym lab	83.3% ✎	▄···▄··▄█▄█▄ beats 7/8	medium · avg 36%	▇ 1.2%	small run
ExploitGym (pass@1) lab	5.4% ✎	▄···▄··▄▄··█ beats 2/5	very hard · avg 0.5%	▇ 1.0%	too hard for it · noisy · small run
AutomationBench (DeepSeek V4.1 card) lab	43.2% ✎	█··█·······█ beats 1/2	very hard · avg 12%	▇ 0.9%	at its level · small run
RobloxQA v2.0 multiple choice (Torpedo Software)	92.6% ✎	▁▁▃▁█▄▃▃▆▂▂▂ beats 23/49	easy · avg 78%	▇ 0.8%	maxed out
AutomationBench (Public) lab	31.8% ✎	█··········█ beats 1/1	very hard · avg 6.5%	▇ 0.8%	at its level · small run
ProgramBench (Almost@1) lab	15.5% ✎	█··········█ beats 0/1		—	not counted yet: no other model in the same run to compare it with

$ where to run it

Through an API

api · Deep Infra · $1.63/M · $1.3 in / $2.6 out per 1M tokens · 1M context · deepinfra.com ↗ · price sourceStandard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ.

llama.cpp · recommended

Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
Windows
irm https://llama.app/install.ps1 | iex
other ways to install ↗
Run DeepSeek V4 Pro (0813) · needs ~960.5 GB · from huggingface.co/unsloth
chat in the terminal
llama cli -hf unsloth/DeepSeek-V4-Pro-0813-GGUF:UD-Q4_K_XL
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf unsloth/DeepSeek-V4-Pro-0813-GGUF:UD-Q4_K_XL

LM Studio

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run DeepSeek V4 Pro (0813) · needs ~960.5 GB · from huggingface.co/unsloth
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/unsloth/DeepSeek-V4-Pro-0813-GGUF@UD-Q4_K_XL -y
Then run lms ls to see its name, and lms chat followed by that name

Weights and documentation: huggingface.co/deepseek-ai ↗

$ how the score was worked out · Epoch AI — internal evaluatio… 46%, deepseek-ai/DeepSeek-V4.1-Fla… 24%, LMArena leaderboard dataset 18% and 5 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
Epoch AI — internal evaluations	7	▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 46%	CC BY 4.0
deepseek-ai/DeepSeek-V4.1-Flash model card · lab-reported	14	▇▇▇▇▇▇▇ 24%	MIT
LMArena leaderboard dataset	3	▇▇▇▇▇ 18%	CC BY 4.0
Roblox LLM Leaderboard (Torpedo Software): RobloxQA v2.0 and Luau Leetcode	2	▇ 3%	MIT
GameLogicBench: Evaluating Coding Agents on Runtime Game Logic with Tick-Level State Assertions (arXiv:2609.21562v2)	1	▇ 3%	CC BY 4.0
NVIDIA ModelOpt quantized-checkpoint cards: baseline (unquantized) measurements of third-party models	1	▇ 3%	Apache-2.0 (Gemma-4-31B-IT-NVFP4 repo) / MIT (DeepSeek-V4-Pro-0813-NVFP4 repo)
deepseek-ai/DeepSeek-V4-Pro-0813 model card · lab-reported	2	▇ 1%	MIT
Miðeind Icelandic LLM Leaderboard	1	▇ 1%	CC BY 4.0
prior · pull toward the middle		▇ 0%
weighted average		= 76.4