DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and...
Tested on 16 benchmarks · BenchGecko score 68.6. Top scores: Chatbot Arena Elo · Overall (1436.0%), LiveBench · Mathematics (79.7%), Dtbench (77.3%).
DeepSeek V4.1 Flash 1017 scores 68.4 (100% as good) at $0.01/1M input · 93% cheaper
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.
Unusual and adversarial machine learning challenges. Tests robustness of reasoning about edge cases in ML systems.
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.
Deceptively simple questions that humans find easy but AI models often get wrong. Tests common sense and reasoning gaps.
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~$0.000
Frequently Asked Questions
Related Models
Claude Instant · AnthropicDeepSeek V4.1 Flash 1017 · DeepSeekMixtral 8x7B Instruct · Mistral AIPaLM 2-M · UnknownClaude Opus 4.8 · AnthropicKey facts · as of 2026-11-11
- DeepSeek V4 Flash by DeepSeek. BenchGecko score 68.6, rank 48 of 326 scored models (normalized average of public benchmark scores).
- List price $0.0900 input · $0.18 output per 1M tokens (as of 2026-11-11).
- Sold by 13 providers (as of 2026-11-11): Relace (fp4) $0.0500 in / $0.45 out · StreamLake (fp8) $0.0826 in / $0.17 out · DeepInfra (fp8) $0.0900 in / $0.18 out · GMICloud (fp8) $0.0910 in / $0.18 out · Venice $0.0966 in / $0.19 out · and 8 more. Every provider
- Gecko Tests · Gecko Score 57 (rank 16): Who Are You B (Knows who made it) · Censorship Index D (Refuses or dodges 5%) · Knowledge Horizon D (Knows news up to Apr 2025) · Tokenizer Tax A (7% more tokens outside English).
How to cite · data as of 2026-11-11
DeepSeek V4 Flash · benchmarks, pricing and providers. BenchGecko, data as of 2026-11-11. https://benchgecko.ai/model/deepseek-v4-flash
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