DeepSeek: DeepSeek V4 Pro

Open MoE flagship with million-token context for coding and long agent runs

Source-linked deepseek-thinking Open weights Released 2026-04-24

API record Report

InputT

OutputT

Input price$0.66/M

Output price$1.98/M

Context1M

Max output384K

Providers72

Inference availability

Providers

Provider-specific identifiers, limits, and listed prices per million tokens. Every row links back to the provider's own documentation.

Report a price

Providers offering DeepSeek V4 Pro
Provider	Provider model ID	Context	Max output	Input	Output	Cache read	Capabilities	Docs
Alibaba Token Plan (China)	`deepseek-v4-pro`	1M	384K	—	—	—	ReasoningToolsJSON	Docs ↗
AnyAPI	`deepseek/deepseek-v4-pro`	1M	384K	—	—	—	ReasoningToolsJSON	Docs ↗
Model Oracle AI	`deepseek-v4-pro`	1M	384K	—	—	—	ReasoningToolsJSON	Docs ↗
Nvidia	`deepseek-ai/deepseek-v4-pro`	1.04858M	393.216K	—	—	—	ReasoningToolsJSON	Docs ↗
UnoRouter	`deepseek-v4-pro:free`	1M	384K	—	—	—	ReasoningToolsJSON	Docs ↗
Alibaba Token Plan	`deepseek-v4-pro`	1M	384K	—	—	—	ReasoningToolsJSON	Docs ↗
Umans AI Coding Plan	`umans-deepseek-v4-pro-0813`	1.04858M	393.215K	—	—	—	ReasoningToolsJSON	Docs ↗
Volcengine Ark Coding Plan	`deepseek-v4-pro`	1M	384K	—	—	—	ReasoningToolsJSON	Docs ↗
SCNet Token Plan	`DeepSeek-V4-Pro`	1M	384K	—	—	—	ReasoningToolsJSON	Docs ↗
SenseNova (China)	`deepseek-v4-pro`	1.04858M	65.536K	—	—	—	ReasoningToolsJSON	Docs ↗
OpenRouter	`deepseek/deepseek-v4-pro`	1.04858M	384K	$0.209	$0.418	$0.017	ReasoningToolsJSON	Docs ↗
routing.run	`deepseek-v4-pro`	1M	64K	$0.348	$0.696	—	ReasoningToolsJSON	Docs ↗
CrofAI	`deepseek-v4-pro`	1M	131.072K	$0.35	$0.8	$0.003	ReasoningToolsJSON	Docs ↗
EBCloud	`DeepSeek-V4-Pro`	1M	384K	$0.429	$0.857	—	ReasoningToolsJSON	Docs ↗
Hugging Face	`deepseek-ai/DeepSeek-V4-Pro`	1.04858M	393.216K	$0.435	$0.87	$0.004	ReasoningToolsJSON	Docs ↗
OpenCode Go	`deepseek-v4-pro`	1M	384K	$0.435	$0.87	$0.004	ReasoningToolsJSON	Docs ↗
Pioneer	`deepseek-ai/DeepSeek-V4-Pro`	1M	131.072K	$0.435	$0.87	$0.004	ReasoningToolsJSON	Docs ↗
DevPass (LLM Gateway)	`deepseek-v4-pro`	1.05M	384K	$0.435	$0.87	$0.004	ReasoningToolsJSON	Docs ↗
Modelis	`deepseek-v4-pro`	1M	384K	$0.435	$0.87	—	ReasoningToolsJSON	Docs ↗
TokenGo	`deepseek/deepseek-v4-pro`	1M	384K	$0.435	$0.87	—	ReasoningToolsJSON	Docs ↗
LLM Gateway	`deepseek/deepseek-v4-pro`	1.05M	393.216K	$0.435	$0.87	$0.004	ReasoningTools	Docs ↗
Kenari	`deepseek-v4-pro`	1M	384K	$0.571	$1.143	$0.006	ReasoningToolsJSON	Docs ↗
Vercel AI Gateway	`deepseek/deepseek-v4-pro`	1M	384K	$0.66	$1.98	$0.022	ReasoningToolsJSON	Docs ↗
ZenMux	`deepseek/deepseek-v4-pro`	1M	384K	$0.66	$1.98	$0.022	ReasoningToolsJSON	Docs ↗
OrcaRouter	`deepseek/deepseek-v4-pro`	1M	384K	$0.66	$1.98	$0.022	ReasoningToolsJSON	Docs ↗
DeepSeek	`deepseek-v4-pro`	1.04858M	393.216K	$0.66	$1.98	$0.022	ReasoningToolsJSON	Docs ↗
ClinePass	`cline-pass/deepseek-v4-pro`	1M	384K	$0.66	$1.98	$0.022	ReasoningToolsJSON	Docs ↗
Eden AI	`deepseek/deepseek-v4-pro`	1M	384K	$0.66	$1.98	$0.022	ReasoningToolsJSON	Docs ↗
Vancine	`deepseek-v4-pro`	1M	384K	$0.66	$1.98	$0.022	ReasoningToolsJSON	Docs ↗
Tempr Gateway	`deepseek/deepseek-v4-pro`	1M	384K	$0.66	$1.98	$0.022	ReasoningToolsJSON	Docs ↗
CommandCode	`deepseek/deepseek-v4-pro`	1M	384K	$0.66	$1.98	$0.022	ReasoningToolsJSON	Docs ↗
NanoGPT	`deepseek/deepseek-v4-pro`	1.04858M	384K	$0.7	$1.4	$0.07	ReasoningToolsJSON	Docs ↗
above.dev	`deepseek-v4-pro`	1M	384K	$0.726	$2.178	$0.024	ReasoningToolsJSON	Docs ↗
UnoRouter	`deepseek-v4-pro`	1M	384K	$0.9	$1.8	—	ReasoningToolsJSON	Docs ↗
AI/ML API	`deepseek/deepseek-v4-pro`	1M	384K	$0.9	$2.7	$0.03	ReasoningToolsJSON	Docs ↗
ai&	`deepseek-ai/deepseek-v4-pro`	1.04858M	384K	$1	$2.5	$0.25	ReasoningToolsJSON	Docs ↗
Neuralwatt	`deepseek-v4-pro`	1.04856M	393.216K	$1	$3	$0.1	ReasoningToolsJSON	Docs ↗
Standard Compute	`deepseek/deepseek-v4-pro`	1M	50K	$1.044	$2.088	$0.209	ReasoningToolsJSON	Docs ↗
Fireworks AI	`accounts/fireworks/models/deepseek-v4-pro`	1M	384K	$1.2	$1.2	$0.6	ReasoningToolsJSON	Docs ↗
Deep Infra	`deepseek-ai/DeepSeek-V4-Pro`	1.04858M	16.384K	$1.3	$2.6	$0.1	ReasoningToolsJSON	Docs ↗
Auriko	`deepseek-v4-pro`	1M	384K	$1.32	$3.96	$0.044	ReasoningToolsJSON	Docs ↗
Merge Gateway	`deepseek/deepseek-v4-pro`	1M	384K	$1.32	$3.96	$0.044	ReasoningToolsJSON	Docs ↗
Ofox	`deepseek/deepseek-v4-pro`	1M	384K	$1.32	$3.96	$0.044	ReasoningToolsJSON	Docs ↗
Umans AI	`umans-deepseek-v4-pro-0813`	1.04858M	393.215K	$1.32	$3.96	$0.044	ReasoningToolsJSON	Docs ↗
Requesty	`deepseek-v4-pro`	1M	131.072K	$1.32	$3.96	$0.044	ReasoningToolsJSON	Docs ↗
CrossModel	`deepseek/deepseek-v4-pro`	1M	384K	$1.35	$4.05	$0.045	ReasoningToolsJSON	Docs ↗
GMI Cloud	`deepseek-ai/DeepSeek-V4-Pro`	1.04858M	384K	$1.392	$2.784	$0.116	ReasoningToolsJSON	Docs ↗
SiliconFlow	`deepseek-ai/DeepSeek-V4-Pro`	1M	384K	$1.502	$3.135	$0.135	ReasoningToolsJSON	Docs ↗
Novita AI	`deepseek/deepseek-v4-pro`	1.04858M	393.216K	$1.6	$3.2	$0.135	ReasoningToolsJSON	Docs ↗
Kilo Gateway	`deepseek/deepseek-v4-pro`	1.024M	384K	$1.6	$3.2	$0.135	ReasoningToolsJSON	Docs ↗
Jalapeno Cloud	`DeepSeek-V4-Pro`	1.04858M	384K	$1.6	$3.38	$0.14	ReasoningToolsJSON	Docs ↗
Alibaba (China)	`deepseek-v4-pro`	1M	384K	$1.65	$3.301	—	ReasoningToolsJSON	Docs ↗
EmpirioLabs AI	`deepseek-v4-pro`	1M	393.216K	$1.65	$3.3	$1.65	ReasoningToolsJSON	Docs ↗
Venice AI	`deepseek-v4-pro`	1M	32.768K	$1.65	$3.301	$0.33	ReasoningToolsJSON	Docs ↗
Cortecs	`deepseek-v4-pro`	1M	384K	$1.7	$3.4	$0.15	ReasoningTools	Docs ↗
Vivgrid	`deepseek-v4-pro`	1M	384K	$1.74	$3.48	$0.14	ReasoningToolsJSON	Docs ↗
DigitalOcean	`deepseek-v4-pro`	1.04858M	393.216K	$1.74	$3.48	—	ReasoningToolsJSON	Docs ↗
Abacus	`deepseek-ai/DeepSeek-V4-Pro`	1M	32.768K	$1.74	$3.48	$0.15	ReasoningToolsJSON	Docs ↗
OpenCode	`deepseek-v4-pro`	1M	384K	$1.74	$3.48	$0.145	ReasoningToolsJSON	Docs ↗
HPC-AI	`deepseek/deepseek-v4-pro`	1.002M	128K	$1.74	$3.48	$0.145	ReasoningToolsJSON	Docs ↗
FastRouter	`deepseek/deepseek-v4-pro`	1M	384K	$1.74	$3.48	$0.145	ReasoningToolsJSON	Docs ↗
Impossibl	`deepseek/deepseek-v4-pro`	1M	384K	$1.74	$3.48	$0.145	ReasoningToolsJSON	Docs ↗
Cloudflare AI Gateway	`deepseek/deepseek-v4-pro`	131.072K	384K	$1.74	$3.48	$0.145	ReasoningToolsJSON	Docs ↗
Arcee	`deepseek/deepseek-v4-pro`	512K	384K	$1.74	$3.48	$0.2	ReasoningToolsJSON	Docs ↗
Baseten	`deepseek-ai/DeepSeek-V4-Pro`	1.04858M	262.144K	$1.74	$3.48	$0.145	ReasoningToolsJSON	Docs ↗
Nebius Token Factory	`deepseek-ai/DeepSeek-V4-Pro`	1.04858M	1.04858M	$1.75	$3.5	$1.75	ReasoningToolsJSON	Docs ↗
TensorX	`deepseek/deepseek-v4-pro`	1.04858M	384K	$1.75	$3.5	$0.438	ReasoningToolsJSON	Docs ↗
Opper	`deepseek-v4-pro`	1M	65.536K	$1.788	$3.576	—	ReasoningToolsJSON	Docs ↗
Moark	`DeepSeek-V4-Pro`	1M	384K	$1.8	$3.6	—	ReasoningToolsJSON	Docs ↗
Melious	`deepseek-v4-pro`	1M	384K	$1.855	$3.709	$0.464	ReasoningToolsJSON	Docs ↗
Prime Inference	`deepseek/deepseek-v4-pro`	1.04858M	384K	$1.91	$3.83	—	ReasoningToolsJSON	Docs ↗
Charm Hyper	`deepseek-v4-pro`	1M	384K	$2.4	$4.8	$0.2	ReasoningToolsJSON	Docs ↗

Capability badges appear only where the provider catalog explicitly lists support. A blank cell means the source is silent, not that the feature is absent.

Listed rates

Price across providers

Input price per million tokens as published by each provider. Bars are drawn from listed rates only — no traffic weighting, since the catalog observes no requests.

Lowest input $0.209/M

Across 62 priced providers

Median input $1.32/M

Midpoint of listed rates

Highest input $2.4/M

11.5× the lowest listed rate

Output range $0.418 – $4.8

Per million output tokens

OpenRouter $0.209/MLowest

routing.run $0.348/M

CrofAI $0.35/M

EBCloud $0.429/M

Hugging Face $0.435/M

OpenCode Go $0.435/M

Pioneer $0.435/M

DevPass (LLM Gateway) $0.435/M

Modelis $0.435/M

TokenGo $0.435/M

LLM Gateway $0.435/M

Kenari $0.571/M

Vercel AI Gateway $0.66/M

ZenMux $0.66/M

OrcaRouter $0.66/M

DeepSeek $0.66/M

ClinePass $0.66/M

Eden AI $0.66/M

Vancine $0.66/M

Tempr Gateway $0.66/M

CommandCode $0.66/M

NanoGPT $0.7/M

above.dev $0.726/M

UnoRouter $0.9/M

AI/ML API $0.9/M

ai& $1/M

Neuralwatt $1/M

Standard Compute $1.044/M

Fireworks AI $1.2/M

Deep Infra $1.3/M

Auriko $1.32/M

Merge Gateway $1.32/M

Ofox $1.32/M

Umans AI $1.32/M

Requesty $1.32/M

CrossModel $1.35/M

GMI Cloud $1.392/M

SiliconFlow $1.502/M

Novita AI $1.6/M

Kilo Gateway $1.6/M

Jalapeno Cloud $1.6/M

Alibaba (China) $1.65/M

EmpirioLabs AI $1.65/M

Venice AI $1.65/M

Cortecs $1.7/M

Vivgrid $1.74/M

DigitalOcean $1.74/M

Abacus $1.74/M

OpenCode $1.74/M

HPC-AI $1.74/M

FastRouter $1.74/M

Impossibl $1.74/M

Cloudflare AI Gateway $1.74/M

Arcee $1.74/M

Baseten $1.74/M

Nebius Token Factory $1.75/M

TensorX $1.75/M

Opper $1.788/M

Moark $1.8/M

Melious $1.855/M

Prime Inference $1.91/M

Charm Hyper $2.4/M

Cost calculator

Estimate a workload

Provider Input tokens 1,000,000 Output tokens 250,000 Cached input 0%

$0.00

Excludes taxes, non-token charges, and tiered discounts.

Context limits also differ by provider, from 131.072K to 1.05M tokens. Compare the provider table above before choosing on price alone.

Specification

Capabilities

Recorded from the source catalog and provider listings.

✓ Reasoning Yes

✓ Tool calling Yes

✓ Structured output Yes

× Attachments No

× Vision input No

✓ Open weights Yes

Creator: DeepSeek
Model family: deepseek-thinking
Knowledge cutoff: 2025-05
License: Not documented
Release date: 2026-04-24
Model ID: deepseek/deepseek-v4-pro

Weights: Hugging Face ↗

Published evaluations

Benchmarks

Every result stays attached to its source, version, metric and harness. Scores from different versions are never merged, and the profile below plots each benchmark against its own population rather than on a shared scale.

Benchmark registry

Benchmark profile

Each axis is this model's percentile among the 8 benchmarks it has published results for, measured against every other model with a score on that same benchmark. Percentiles are used because benchmarks do not share a scale — a 60 on one is not a 60 on another. Hover any point for the raw score.

Intelligence Index index Source ↗

Intelligence Index #94 of 334

30.4

Coding Index index Source ↗

Coding Index #69 of 214

59.4

Agentic Index index Source ↗

Agentic Index #74 of 201

26.3

Design Arena: website elo Source ↗

Design Arena: website #86 of 200

1234.0

Design Arena: gamedev elo Source ↗

Design Arena: gamedev #76 of 191

1229.0

Design Arena: codecategories elo Source ↗

Design Arena: codecategories #79 of 188

1240.0

Design Arena: dataviz elo Source ↗

Design Arena: dataviz #92 of 183

1203.0

Design Arena: uicomponent elo Source ↗

Design Arena: uicomponent #89 of 182

1216.0

Design Arena: 3d elo Source ↗

Design Arena: 3d #54 of 176

1254.0

Design Arena: svg elo Source ↗

Design Arena: svg #83 of 128

1140.0

Design Arena: asciiart elo Source ↗

Design Arena: asciiart #69 of 108

1149.0

Design Arena: webapps elo Source ↗

Design Arena: webapps #80 of 80

1000.0

Design Arena: fullstack elo Source ↗

Design Arena: fullstack #69 of 69

948.0

Design Arena: godotgamedev elo Source ↗

Design Arena: godotgamedev #50 of 55

1059.0

SWE-Bench Verified resolved Source ↗

SWE-Bench Verified #5 of 32

80.6

Terminal-Bench — 2.1 pass@1 · Claude Code Source ↗

Terminal-Bench #5 of 7

64.7

Artificial Analysis Coding Agent Index average pass@1 · Claude Code Source ↗

Artificial Analysis Coding Agent Index #5 of 6

50.1

GDPval-AA Elo Source ↗

GDPval-AA #3 of 6

1554.0

SWE-Atlas Codebase QnA pass@1 · Claude Code Source ↗

SWE-Atlas Codebase QnA #5 of 6

67.8

SWE-Bench Pro pass@1 · Claude Code Source ↗

SWE-Bench Pro #4 of 6

18.0

Terminal-Bench — 2.0 accuracy Source ↗

Terminal-Bench #1 of 4

67.9

Codeforces rating Source ↗

Codeforces #2 of 3

3206.0

GPQA Diamond pass@1 Source ↗

GPQA Diamond #2 of 3

90.1

Humanity's Last Exam pass@1 Source ↗

Humanity's Last Exam #2 of 3

48.2

MathArena Apex pass@1 Source ↗

MathArena Apex #2 of 3

38.3

BrowseComp pass@1 Source ↗

83.4 2 models scored — too few for a distribution

Chinese SimpleQA pass@1 Source ↗

84.4 2 models scored — too few for a distribution

CorpusQA accuracy Source ↗

62.0 2 models scored — too few for a distribution

HMMT pass@1 Source ↗

95.2 2 models scored — too few for a distribution

IMOAnswerBench pass@1 Source ↗

89.8 2 models scored — too few for a distribution

LiveCodeBench pass@1 Source ↗

93.5 2 models scored — too few for a distribution

MCP Atlas pass@1 Source ↗

73.6 2 models scored — too few for a distribution

MMLU-Pro EM Source ↗

87.5 2 models scored — too few for a distribution

MRCR MMR Source ↗

83.5 2 models scored — too few for a distribution

MathArena Apex Shortlist pass@1 Source ↗

90.2 2 models scored — too few for a distribution

SWE-Bench Multilingual resolved Source ↗

76.2 2 models scored — too few for a distribution

SWE-Bench Pro resolved Source ↗

55.4 2 models scored — too few for a distribution

SimpleQA-Verified pass@1 Source ↗

57.9 2 models scored — too few for a distribution

Toolathlon pass@1 Source ↗

51.8 2 models scored — too few for a distribution

Each strip shows every published score for that benchmark, with this model marked. The lighter shape behind the ticks is the density of results, and the dashed line is the median.

Cost against capability

Price and performance

Listed input price plotted against Intelligence Index, the benchmark with the widest published coverage that this model appears in.

The stepped line is the efficient frontier: at each price, the best score available for that money or less. A model sitting on it is not being beaten by anything cheaper. Price is log-scaled because listed rates span four orders of magnitude. Only models with both a listed price and a score on this benchmark can appear.

Catalog activity

Change log

Field-level changes detected between successful source imports.

Full change log

Oct 24, 2026Context Length1024000 → 1000000

Oct 24, 2026Price Completion0.41759999999999997 → 1.98

Oct 24, 2026Price Prompt0.20879999999999999 → 0.66

Oct 24, 2026Context Length1000000 → 1024000

Oct 24, 2026Price Completion1.98 → 0.41759999999999997

Oct 24, 2026Price Prompt0.66 → 0.20879999999999999

Oct 24, 2026Context Length1024000 → 1000000

Oct 24, 2026Price Completion0.41759999999999997 → 1.98

Oct 24, 2026Price Prompt0.20879999999999999 → 0.66

Oct 24, 2026Context Length1000000 → 1024000

Provenance

Sources & verification

Every figure on this page traces back to one of these records.

Methodology

Last verified: Oct 24, 2026
Status: Source-linked
Confidence: Medium
Catalog source: Models.dev

Public API

Use this record

Fetch the complete source-linked model record. No key, no account, no rate-limited tier.

API documentation

Endpoint

GET https://modelbench.lol/api/v1/models/deepseek/deepseek-v4-pro

curl

curl "https://modelbench.lol/api/v1/models/deepseek/deepseek-v4-pro"

Common questions

Frequently asked questions

Answered directly from the stored record — nothing here is generated beyond the catalog's own fields.

What is DeepSeek V4 Pro?

Open MoE flagship with million-token context for coding and long agent runs. It is published by DeepSeek and catalogued here from Models.dev.

How much does DeepSeek V4 Pro cost?

Listed input pricing starts at $0.209 per million tokens from OpenRouter, rising to $2.4 across 62 listed providers.

What is the context length of DeepSeek V4 Pro?

DeepSeek V4 Pro accepts up to 1M tokens of context and returns up to 384K output tokens.

Does DeepSeek V4 Pro support tool calling and structured output?

Provider catalogs list support for tool calling, structured output, and reasoning.

Which providers serve DeepSeek V4 Pro?

71 providers list this model: Alibaba Token Plan (China), AnyAPI, Model Oracle AI, Nvidia, UnoRouter, Alibaba Token Plan and 65 more.

Are the weights for DeepSeek V4 Pro open?

Yes. The weights are published and downloadable from Hugging Face.

When was DeepSeek V4 Pro released?

The catalog records a release date of 2026-04-24, last verified Oct 24, 2026.

More models from DeepSeek

View all →

DeepSeek V4 Flash

Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work

1M ctx $0.15/M in DeepSeek V3.2

Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use

128K ctx $0.209/M in DeepSeek: DeepSeek V3.2 Exp

DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...

163.84K ctx $0.27/M in DeepSeek: DeepSeek V3.1 Terminus

DeepSeek-V3.1 Terminus is an update to [DeepSeek V3.1](/deepseek/deepseek-chat-v3.1) that maintains the model's original capabilities while addressing issues reported by users, including language consistency and agent capabilities, further optimizing the model's...

163.84K ctx $0.27/M in