0

DeepSeek V4 Pro

Open weights

DeepSeek's April 2026 next-gen open-weights flagship - 1.6T-total / 49B-active MoE with 1M context and DeepSeek Sparse Attention.

Open
Publisher
DeepSeek
Family
Deepseek
Params
1.6T total / 49B active
Context
1.0M
Input
Output
Providers
deepseektogetherfireworksself-host
License
mit
Released
24 Apr 2026
Canonical
api-docs.deepseek.com/news/news260424

Cite

Notes

Only stored in your browser.

Attribution

Benchmark scores
AALiveBenchVaultmatharenaVals AI
Attribution policy →
Context
1.0M
Input
$0.43
/1M tok
Output
$0.87
/1M tok
Blended
$0.54
#267/ 414/1M · 3:1
Speed
55
#165/ 414tok/s
Cache read
$0
0.01x input

Price & speed via Artificial Analysis · ranked across tracked models · caching via OpenRouter

Reported on 23 evals across 10 domains - best 91.2% on τ²-bench (Tau²-bench) · #32 of 320

Reported eval scores

23
SWE-bench80.6
LiveBench - Math90.7
LiveBench - Language78.1
Terminal-Bench (Hard)36.4
LiveBench - Reasoning82.7
LiveBench - Data Analysis74.5
LiveBench - Coding70
IFBench45.8
LiveBench - Instruction Following62.4
SciCode42.4
GPQA Diamond71.7
τ²-bench (Tau²-bench)91.2
LiveBench76.4
Vals Index56.2
Humanity's Last Exam (HLE)7.7
Vibe Code Bench v1.149.9
MathArena76.5
CorpFin v261.4
TaxEval v272.1
Finance Agent (v2)44.1
MedCode40.5
MedScribe75.1
ProofBench10