DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with 1.6T total parameters and 49B activated parameters, supporting a 1M-token context window. It is designed for advanced reasoning, coding,...
Tested on 34 benchmarks · BenchGecko score 60.8. Top scores: Chatbot Arena Elo — Overall (1457.8%), Chatbot Arena Elo — Coding (1446.0%), OTIS Mock AIME 2024-2025 (96.7%).
gpt-oss-20b (free) scores 61.0 (100% as good) at $0.00/1M input · 100% cheaper
Real-world software engineering tasks from GitHub issues. Models must diagnose bugs and write patches that pass test suites. Human-verified subset of SWE-bench.
Regularly refreshed coding problems that avoid data contamination. New problems added monthly to prevent memorization.
LiveBench coding tasks that require multi-step reasoning and tool use. Tests planning and execution of complex coding workflows.
Regularly refreshed reasoning problems testing logical deduction, spatial reasoning, and analytical thinking.
Fresh data analysis tasks testing ability to interpret tables, charts, and statistical data.
Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.
Regularly updated math problems that test numerical reasoning, algebra, calculus, and combinatorics.
- Typetext
- Context1.0M tokens (~524 books)
- ReleasedApr 2026
- LicenseOpen Source
- StatusActive
- Cost / Message~$0.001
Frequently Asked Questions
Related Models
o3 Pro · OpenAIGemini 3.6 Flash · Google DeepMindgpt-oss-20b (free) · OpenAIGPT-5 · OpenAIGemini 3.5 Flash · Google DeepMindKey facts · as of 2026-10-05
- DeepSeek V4 Pro by DeepSeek. BenchGecko score 60.8, rank 81 of 312 scored models (normalized average of public benchmark scores).
- List price $0.21 input · $0.42 output per 1M tokens (as of 2026-10-05).
- Sold by 16 providers (as of 2026-10-05): Relace (fp4) $0.21 in / $4.20 out · StreamLake (fp8) $0.21 in / $0.42 out · Parasail (fp8) $0.45 in / $3.48 out · Reka $0.81 in / $2.08 out · GMICloud (fp8) $0.96 in / $1.91 out · and 11 more. Every provider
- Gecko Tests: Who Are You B (Knows who made it) · Tokenizer Tax D (68% more tokens outside English).
How to cite · data as of 2026-10-05
DeepSeek V4 Pro · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-05. https://benchgecko.ai/model/deepseek-v4-pro
Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP