DeepSeekOpen weightsReasoning modelCompare with another model

DeepSeek V4 Pro 0424

Reasoning effort

DeepSeek V4 Pro 0424 is a DeepSeek open-weights reasoning model, released 24 Apr 2026. Its best configuration (max reasoning effort) ranks #125 of 784 on the BenchLeader Index at 59.7 ±4.8, in the upper half. The ± is the point: 234 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (75) and lowest in knowledge (57). At $0.548 per million tokens blended it is mid-priced. Output speed of 92 tokens per second puts it faster than most, with a first answer in 49.0 s. It has been measured at 3 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 24 Apr 2026.stale: no new result in six months

Blended price: $0.548/M; $0.440 in · $0.870 out
Output speed: 92 tok/s; measured by Artificial Analysis
First answer: 49 s
Context: 1M
Full answer: 54 s; median, reasoning included
Cost per run: $0.122; one full Intelligence Index run
Released: 24 Apr 2026

How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.

Overall index60
Reasoning62
Agents & tools64
Knowledge57
Instruction following75
Long context62
Composite64

Versions

DeepSeek has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.

Model	Released	Index	Rank
DeepSeek V4 Promax	13 Aug 2026	60.8	#102
DeepSeek V4 Pro 0424maxthis page	24 Apr 2026	59.7	#125

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.

Effort	Index	Rank	Speed	First answer	Chat reply cost	Categories
no reasoning	50.8	#359	92 tok/s	1.67 s	$0.0004	Agents & tools 66 · Composite 53 · Instruction following 48 · Knowledge 48 · Long context 51 · Reasoning 42
high	57.9	#167	86 tok/s	25 s	$0.0004	Agents & tools 60 · Composite 63 · Instruction following 70 · Knowledge 57 · Long context 60 · Reasoning 57
maxbest	59.7	#125	92 tok/s	49 s	$0.0004	Agents & tools 64 · Composite 64 · Instruction following 75 · Knowledge 57 · Long context 62 · Reasoning 62

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmark	no reasoning	high	max	Source
GPQA Diamond (AA)not in index	71.7%#284	90.5%#59	88.8%#86	Artificial Analysis
Humanity's Last Exam (AA)not in index	8.3%#350	35.2%#117	37.5%#102	Artificial Analysis
CritPt	0.9%#236	10.0%#120	12.9%#104	Artificial Analysis

Coding

Benchmark	high	max	Source
SciCode (AA)not in index	–	50.8%#101	Artificial Analysis
Terminal-Bench 4.0 (AA)not in index	–	14.7%#70	Artificial Analysis
Terminal-Bench 2.1 (AA)not in index	64.8%#85	64.0%#87	Artificial Analysis

Agents & tools

Benchmark	no reasoning	high	max	Source
Terminal-Bench Hard	36.4%#71	41.7%#50	46.2%#30	Artificial Analysis
τ²-Bench Telecom (AA)not in index	91.2%#59	94.2%#33	96.2%#14	Artificial Analysis
GDPval-AA v2.1not in index	–	32.6%#130	32.8%#129	Artificial Analysis
τ³-Banking (AA)	–	26.2%#78	30.1%#66	Artificial Analysis
ITBench SRE (AA)not in index	–	–	38.3%#22	Artificial Analysis
Analyst Agent (AA)not in index	–	–	18.8%#37	Artificial Analysis
APEX-Agents (AA)not in index	–	–	24.3%#19	Artificial Analysis
AutomationBenchnot in index	–	–	56.3%#53	Artificial Analysis
GDP.pdf (AA)not in index	–	–	13.4%#104	Artificial Analysis
EnterpriseOps-Gymnot in index	–	–	40.4%#28	Artificial Analysis

Knowledge

Benchmark	no reasoning	high	max	Source
AA-Omniscience	-29.9#279	-10.6#196	-10.7#197	Artificial Analysis
AA-Omniscience: accuracynot in index	30.9%#174	41.4%#113	43.0%#103	Artificial Analysis
AA-Omniscience: non-hallucinationnot in index	12.2%#378	11.3%#390	5.9%#511	Artificial Analysis

Instruction following

Benchmark	no reasoning	high	max	Source	Trend
IFBench	45.8%#199	71.3%#57	76.5%#20	Artificial Analysis

Long context

Benchmark	no reasoning	high	max	Source	Trend
AA-LCR	53.0%#308	70.3%#210	74.7%#158	Artificial Analysis
MLCRnot in index	–	–	9.4%#114	Artificial Analysis

Composite

Benchmark	no reasoning	high	max	Source	Trend
AA Intelligence Index v4.3.2	20.8#218	30.1#121	30.4#118	Artificial Analysis

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0004	–	52.2 s
Summarize a 30-page report	12,000 / 600	$0.0058	–	55.5 s
Code edit	6,000 / 1,500	$0.0039	–	1.1 min
Agentic coding session	60,000 / 4,000	$0.030	–	1.5 min
Structured extraction	2,000 / 200	$0.0011	–	51.1 s