DeepSeek V4.1 Flash 1017 Thinking

The October 17, 2026 snapshot of DeepSeek V4.1 Flash supports text and image input, tool calling, and optional reasoning with low, high, and max effort. It has a shared 1,048,576-token context window and supports up to 393,216 output tokens within the remaining context. Released under the MIT license.

Reasoning
Vision
Tool Calling

Added Oct 17, 2026

Model weights

Pricing

Auto routing · per 1M tokens

Input: $0.100
Output: $0.60
Cache read: $0.0030

Compare provider prices

Specifications

Context window: 1M
Max output: 393.2K
Avg output (7d): 1.3K tokens; Longer than 78% of models

Benchmarks

Sourced from Artificial Analysis.

Intelligence Index

45.1

Better than 96% of models compared

Agentic work

AutomationBench-AA

Workflow automation with guardrail penalties

66.6%

Better than 95% of models compared

AutomationBench-AA Tasks Completed

Fully completed workflows without guardrail violations

36.7%

Better than 93% of models compared

AA-Briefcase v1.1

Agentic knowledge work (Elo)

1570 Elo

Better than 91% of models compared

GDPval-AA v2.1

Economically valuable tasks (Elo)

1746 Elo

Better than 98% of models compared

Document reasoning

GDP.pdf

Professional PDF reasoning: all-pass rate

17.2%

Better than 63% of models compared

AA-LCR v1.1

Long context reasoning with updated grading

82.7%

Better than 94% of models compared

Reasoning

HLE

Humanity's Last Exam

43.3%

Better than 91% of models compared

CritPt

Research-level physics reasoning

14.3%

Coding

Terminal-Bench v4.0

Practical coding and terminal tasks

33.8%

Better than 83% of models compared

SciCode

Python programming for scientific computing

55.9%

Better than 82% of models compared

Knowledge

AA-Omniscience Accuracy

Proportion of correctly answered questions

46.4%

AA-Omniscience Hallucination Rate

Rate of incorrect answers among non-correct responses

96.5%

Legacy benchmarks

AA-LCR (unversioned / legacy)

Long context reasoning evaluation

82.7%

Better than 94% of models compared

GDPval-AA (unversioned / legacy)

Economically valuable tasks

55.0%

Last updated Oct 22, 2026

Artificial Analysis

Providers

Choose explicit providers for this model. Auto routing remains available as the default option.

Loading provider options…