DeepSeek

🇨🇳ChinaWebsite

Top Model

DeepSeek V3.2 Speciale

78.2

avg score

Compare DeepSeek models All providers

Total Models

tracked on BenchGecko

Open Source

93% of models

$0.09

Cheapest Model

per 1M input tokens

43.3

Avg Benchmark

across 23 scored models

Model Categories

LLM28

Pricing Range — $/1M input tokens

$0.09

$0.20

$0.21

$0.23

$0.27

$0.29

$0.40

$0.43

$0.50

$0.70

$0.80

Low: $0.09Median: $0.27High: $0.80

Open Source Ratio

93%

26 open source2 proprietary

All DeepSeek Models28 total

#▲	Model	Avg	aider edit	aider poly?	ANLI?	APEX-Agents?	ARC AI2?	ARC-AGI?	ARC-AGI-2?	aa agentic?	aa coding ?	aa quality	seal audio	seal audio	seal audio	Balrog?	BBH?	hf bbh	C-Eval	CadEval?	charxiv re?	charxiv re?	arena elo	arena elo	chess puzz?	CMMLU	CSQA2	Cybench?	deepresear?	EnigmaEval	fiction li?	Fortress	frontierma?	frontierma?	GeoBench?	GPQA	GPQA diamond?	graphwalks?	GSM8K?	GSO-Bench?	HellaSwag?	HELM · GPQA	helm ifeva	helm mmlu	helm omni	helm wildb	HLE?	hle tools	seal human	seal human	IFEval	jp jcommon	JHumanEval	JMMLU	JNLI	JSQuAD	LAMBADA?	lech mazur?	livebench	livebench	livebench	livebench	livebench	livebench	livebench	livebench	jp overall	MASK	MATH level 5?	MATH Level 5	MCP Atlas	MMLU?	MMLU-PRO	MMMLU	mmmlu ar	mmmlu bn	mmmlu zh	mmmlu fr	mmmlu de	mmmlu hi	mmmlu id	mmmlu it	mmmlu ja	mmmlu ko	mmmlu pt	mmmlu es	mmmlu sw	mmmlu yo	seal multi	MultiNRC	MUSR	OpenBookQA?	oc aime202	oc gpqa di	oc hle	oc ifeval	oc livecod	oc mmlu pr	OSWorld?	otis mock ?	PIQA?	posttrainb	seal pro r	seal pro r	seal prope	seal remot	ScienceQA?	SciPredict	SimpleBench?	simpleqa v?	seal swe a	seal swe a	swe bench	swe bench	swe bench	seal swe b	seal swe b	swe bench ?	swe bench ?	terminal b?	the agent ?	TriviaQA?	TutorBench	USAMO	VideoMME?	VISTA	seal visua	VPCT?	WeirdML?	Winogrande?	$/1M in	Context	Released
1	DeepSeek V3.2 Speciale🇨🇳 DeepSeekOpen	95.2	-	-	-	-	-	-	-	0.0	37.9	29.4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	96.0	86.7	28.6	91.7	80.9	85.5	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	$0.40	164K	Dec 257mo ago
2	DeepSeek R1 Distill Qwen 14B🇨🇳 DeepSeekOpen	88.3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	40.7	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	18.3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	43.8	93.7	-	63.4	82.4	89.8	-	-	-	-	-	-	-	-	-	-	56.8	-	-	57.0	-	-	40.7	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	28.7	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	N/A	-	Jan 251y ago
3	DeepSeek-V2 (MoE-236B, May 2024)🇨🇳 DeepSeekOpen	84.4	-	-	-	-	89.6	-	-	-	-	-	-	-	-	-	71.7	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	82.8	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	71.2	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	67.8	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	80.0	-	-	-	-	-	-	-	72.6	N/A	-	Jan 242y ago
4	DeepSeek V3.2🇨🇳 DeepSeekOpen	58.7	-	74.2	-	-	-	57.0	4.0	52.9	36.7	41.7	-	-	-	-	-	-	-	-	-	-	1326.9	1424.4	14.0	-	-	-	-	-	-	-	22.1	2.1	-	-	77.9	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	46.7	75.7	45.0	23.1	64.2	64.0	51.8	44.3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	93.0	84.6	23.2	89.7	75.4	85.8	-	87.8	-	-	-	-	-	-	-	-	-	27.5	-	-	-	-	-	-	-	-	-	39.6	-	-	-	-	-	-	-	-	-	-	$0.25	131K	Dec 257mo ago
5	DeepSeek V3.2 Exp🇨🇳 DeepSeekOpen	58.7	-	74.2	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	1285.5	1422.8	-	-	-	-	-	-	83.3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	36.7	73.2	44.3	19.3	65.6	64.4	49.9	45.5	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	42.9	-	-	-	-	-	-	-	39.5	-	$0.27	164K	Sep 259mo ago
6	DeepSeek V3🇨🇳 DeepSeekOpen	58.3	-	48.4	-	-	93.7	-	-	-	-	-	-	-	-	-	83.3	-	-	-	-	-	-	1358.2	-	-	-	-	-	-	50.0	-	1.7	-	-	-	42.0	-	-	-	85.2	53.8	83.2	72.3	40.3	83.1	-	-	-	-	-	-	-	-	-	-	-	77.0	-	-	-	-	-	-	-	-	-	-	64.8	-	-	82.9	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	15.8	69.4	-	-	-	-	-	-	-	2.7	-	-	-	-	-	-	-	-	-	-	-	-	82.9	-	-	-	-	-	-	36.1	70.4	$0.32	164K	Dec 241y ago
7	R1 0528🇨🇳 DeepSeekOpen	56.4	-	71.4	-	-	-	21.2	1.1	20.8	24.0	27.1	-	-	-	-	-	-	-	-	-	-	-	1421.7	-	-	-	-	35.1	-	-	-	-	-	-	-	68.4	-	-	-	-	66.6	78.4	79.3	42.4	82.8	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	96.6	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	89.0	80.6	14.4	80.0	61.0	83.5	-	66.4	-	-	-	-	-	-	-	-	29.0	27.4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	41.6	-	$0.50	164K	May 251y ago
8	DeepSeek V3.1🇨🇳 DeepSeekOpen	53.4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	1417.9	-	-	-	-	-	-	52.8	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	85.2	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	28.0	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	38.4	-	$0.15	33K	Aug 2510mo ago
9	R1🇨🇳 DeepSeekOpen	48.0	-	56.9	-	-	-	15.8	1.3	-	-	-	-	-	-	34.9	-	-	-	-	-	-	-	1397.5	-	-	-	-	35.1	-	69.4	-	-	-	-	-	62.3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	83.0	-	-	-	-	-	-	-	-	-	-	93.0	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	53.3	-	-	-	-	-	-	-	-	17.1	27.4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	36.5	-	$0.70	64K	Jan 251y ago
10	R1 Distill Llama 70B🇨🇳 DeepSeekOpen	47.6	-	-	-	-	-	-	-	-	11.4	15.9	-	-	-	-	-	35.8	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	2.0	-	-	-	-	-	-	-	-	-	-	-	-	-	-	43.4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	30.7	-	-	41.6	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	13.3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	$0.70	131K	Jan 251y ago
11	R1 Distill Qwen 32B🇨🇳 DeepSeekOpen	42.2	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	17.1	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	4.6	-	-	-	-	-	-	-	-	-	-	-	-	-	-	41.9	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	17.1	-	-	41.0	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	16.1	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	$0.29	33K	Jan 251y ago
12	DeepSeek R1 Distill Llama 8B🇨🇳 DeepSeekOpen	38.4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	5.3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	0.7	-	-	-	-	-	-	-	-	-	-	-	-	-	-	37.8	62.4	-	37.8	69.4	80.2	-	-	-	-	-	-	-	-	-	-	41.4	-	-	22.0	-	-	12.1	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	0.5	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	N/A	-	Jan 251y ago
13	DeepSeek R1 Distill Qwen 7B🇨🇳 DeepSeekOpen	38.3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	7.9	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	3.9	-	-	-	-	-	-	-	-	-	-	-	-	-	-	40.4	59.8	-	42.3	54.6	74.2	-	-	-	-	-	-	-	-	-	-	39.3	-	-	19.6	-	-	14.7	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	3.5	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	N/A	-	Jan 251y ago
14	DeepSeek Coder 33B🇨🇳 DeepSeekOpen	26.5	-	-	-	-	22.9	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	35.4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	19.2	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	24.0	N/A	-	Jan 242y ago
15	DeepSeek Coder 6.7B🇨🇳 DeepSeekOpen	16.1	-	-	-	-	15.2	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	21.3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	15.2	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	15.2	N/A	-	Jan 242y ago
16	DeepSeek R1 Distill Qwen 1.5B🇨🇳 DeepSeekOpen	13.9	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	4.7	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	0.8	-	-	-	-	-	-	-	-	-	-	-	-	-	-	34.6	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	16.9	-	-	2.1	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	3.0	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	N/A	-	Jan 251y ago

90+ Gold 80-89 70-79 60-69 <60Scores in % unless noted. Avg = unweighted mean across tested benchmarks.

About DeepSeek

Quick answers · sourced from our data

How many models does DeepSeek have?

BenchGecko tracks 28 models from DeepSeek, of which 26 (93%) are open source. Every entry is updated daily from live provider feeds.

What is the best model from DeepSeek?

DeepSeek V3.2 Speciale is currently the highest scoring DeepSeek model we track, with an average benchmark score of 78.2. Scores are computed across every public benchmark we have data for.

What is the cheapest DeepSeek model?

The cheapest DeepSeek model on BenchGecko starts at $0.09 per 1M input tokens. Pricing is pulled from OpenRouter and cross-checked against official provider rate cards.

How does DeepSeek compare on benchmarks?

DeepSeek models average 43.3 across the benchmarks we track · see the All Providers page for the full ranking by model count, open source ratio, and average score.

Where is DeepSeek based?

DeepSeek is headquartered in China. BenchGecko groups providers by region to make it easy to compare US, EU, China, and Rest of World markets.

Is DeepSeek open source?

26 of 28 DeepSeek models are open source (93%). The rest are proprietary · closed weights served via API.

All Providers 🇨🇳 View by Country DeepSeek · Economy →Pricing →Compare →Full Leaderboard →