Private AI
DeepSeek V4 Flash 0731 Thinking enables reasoning by default on the re-post-trained Mixture-of-Experts model with a 1M-token context window, built for coding, reasoning, and agent workflows. ⚠️ This route goes directly to DeepSeek, so privacy and logging guarantees are limited.
Added Aug 1, 2026
Model weightsContext Window
1.0M
Max Output
384.0K
Input Price (Auto)
$0.094/1M
Output Price (Auto)
$0.19/1M
Cache Read (Auto)
$0.019/1M
Capabilities
Performance metrics and benchmarks
Sourced from Artificial Analysis.
Intelligence Index
49.9
Choose explicit providers for this model. Auto routing remains available as the default option.
Loading provider options…
Coding Index
69.1
Agentic Index
45.7
GPQA Diamond
Graduate-level scientific reasoning
90.8%
Better than 94% of models compared
HLE
Humanity's Last Exam
36.8%
Better than 94% of models compared
AA-LCR
Long context reasoning evaluation
65.7%
Better than 82% of models compared
GDPval-AA
Economically valuable tasks
52.9%
CritPt
Research-level physics reasoning
16.6%
SciCode
Python programming for scientific computing
49.9%
Better than 91% of models compared
AA-Omniscience Accuracy
Proportion of correctly answered questions
37.2%
AA-Omniscience Hallucination Rate
Rate of incorrect answers among non-correct responses
84.4%
Last updated Aug 1, 2026
Artificial Analysis