NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/DeepSeek V4.1 Flash
DeepSeek logo

DeepSeek V4.1 Flash

by DeepSeek · Released Sep 2026

Open SourceMultimodal1M Context
58.6
avg score
Rank #99
Compare
Better than 69% of all models
Context
1.0M tokens (~524 books)
Input $/1M
$0.02
Output $/1M
$1.00
Type
multimodal
License
Open Source
Benchmarks
24 tested
Data as of
About

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...

Tested on 24 benchmarks · BenchGecko score 58.6. Top scores: OTIS Mock AIME 2024-2025 (98.3%), ARC-AGI (94.5%), GPQA diamond (90.0%).

Capabilities
coding
43.9
#128 globally
reasoning
83.7
#16 globally
math
65.4
#68 globally
knowledge
47.8
#152 globally
agentic
39.5
#41 globally
speed
51.6
#27 globally
general
47.0
#43 globally
Benchmark Scores
Compare All
Tested on 24 benchmarks · Ranked across 7 categories
Score Distribution (all 320 models)
0255075100
▲ You are here
codingCompare coding →
Frontiercode
43.9·
reasoningCompare reasoning →
ARC-AGI

Abstraction and Reasoning Corpus. Tests fluid intelligence through novel visual pattern recognition puzzles. Core measure of general intelligence.

94.5·
ARC-AGI-2

ARC-AGI 2, harder sequel to ARC. More complex abstract reasoning patterns that test generalization ability beyond training data.

72.9·
mathCompare math →
OTIS Mock AIME 2024-2025

Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.

98.3·
FrontierMath-Tiers-1-3-v2-Private
70.9·
FrontierMath-Tier-4-v2-Private
26.8·
Quick compare:DeepSeek logovs DeepSeek R1 Distill Qwen 32BGoogle DeepMind logovs Gemini 2.5 ProxAI logovs Grok 4
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · DeepSeek DeepSeek V4.1
Sep 2026
58.6
$0.04/M in1.0Mctx23 benchmarks
Sep 2026
$0.11/M in(+0.08)1.0Mctx
Oct 2026
$0.30/M in(+0.19)1.0Mctx
See the full DeepSeek V4.1 family →
Recently Happened
DeepSeek V4.1 Flash pricing dropped 93%
Oct 25, 2026
DeepSeek V4.1 Flash added
Sep 11, 2026
Similar Models
DeepSeek logo
DeepSeek R1 Distill Qwen 32B
DeepSeek
58.7n/a
Google DeepMind logo
Gemini 2.5 Pro
Google DeepMind
58.5$1.25/1M
xAI logo
Grok 4
xAI
58.5n/a
Links
Info
DeepSeekPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@DeepSeek
Source Code
GitHubHugging Face
BenchGecko API
deepseek-v4-1-flash
Specifications
  • Typemultimodal
  • Context1.0M tokens (~524 books)
  • ReleasedSep 2026
  • LicenseOpen Source
  • StatusActive
  • Cost / Message~$0.001
Available On
DeepSeek logoDeepSeek$0.02
Categories
codingreasoningmathknowledgeagenticspeedgeneral
Learn More
context-windowtransformeropen weightstokens
Share & Export
Tweet
Related Models
DeepSeek logoDeepSeek R1 Distill Qwen 32B
58.7
Google DeepMind logoGemini 2.5 Pro
58.5
xAI logoGrok 4
58.5
Nemotron-4 15B
58.9
DeepSeek V4.1 Flash is an open-source multimodal AI model by DeepSeek, released in September 2026. It has an average benchmark score of 58.6. Context window: 1M tokens.

Related Models

DeepSeek R1 Distill Qwen 32B · DeepSeekGemini 2.5 Pro · Google DeepMindGrok 4 · xAINemotron-4 15B · UnknownGPT-4o (2024-05-13) · OpenAI

Benchmarks

OTIS Mock AIME 2024-2025ARC-AGIGPQA diamondArtificial Analysis · Long Context ReasoningDtbench

Related Pages

DeepSeek · ProviderDeepSeek · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts · as of 2026-10-26

  • DeepSeek V4.1 Flash by DeepSeek. BenchGecko score 58.6, rank 99 of 320 scored models (normalized average of public benchmark scores).
  • List price $0.0200 input · $1.00 output per 1M tokens (as of 2026-10-26).
  • Sold by 41 providers (as of 2026-10-26): OpenInference (fp4) $0.0200 in / $1.00 out · Relace $0.0700 in / $0.50 out · Decart (fp4) $0.0900 in / $0.18 out · Sail Research (fp4) $0.0900 in / $0.25 out · Sail Research (fp4) $0.0900 in / $0.25 out · and 36 more. Every provider
  • Gecko Tests · Gecko Score 32 (rank 38): Who Are You E (Sometimes says it is OpenAI) · Censorship Index E (Refuses or dodges 6%) · Knowledge Horizon A (Knows news up to May 2026) · Tokenizer Tax D (68% more tokens outside English).

How to cite · data as of 2026-10-26

DeepSeek V4.1 Flash · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-26. https://benchgecko.ai/model/deepseek-v4-1-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