Skip to content

deepseek-v4-flash-vision-exp

DeepSeek

legacy model Its test configurations are retired, so it is no longer run. The figures are from its best legacy runs, scored as they ran at the time; it is not ranked.

average±3.4 (95%): 72.2
benchmarksof 12: 12
per 1,000 documentsmean over its benchmarks: $12
seconds per inputmean response time: 72.2

Compare with other models Its row in the Matrix All its test runs

By task

Data correction

68.1

Best of 114 models with all its benchmarks: claude-opus-5-5, 82.7.
Document understanding

74.1

Best of 107 models with all its benchmarks: gpt-6-astra, 97.0.
Information extraction

81.7

Best of 107 models with all its benchmarks: grok-4.6, 90.8.
Transcription

59.6

Best of 114 models with all its benchmarks: gemini-4-argon, 96.2.

By benchmark

Benchmark	Score	Rank	Per 1,000 docs	s / input	Best run
Bibliographic Data	98.1±1.5	–	$15$0.074 run	93.41	T1750 · 8 Sep 2026prompt.txt
Blacklist Cards	94.2±2.2	–	$2.50$0.082 run	15.54	T1757 · 8 Sep 2026prompt.txt
Book Advert XML Repair	49.0±12.1	–	$13$0.64 run	78.15	T1760 · 8 Sep 2026prompt.txt
Business Letters	53.0±7.6	–	$2.35$0.13 run	14.57	T1751 · 8 Sep 2026prompt.txt
Company Lists	90.9±4.6	–	$7.99$0.12 run	49.79	T1758 · 8 Sep 2026prompt.txt
Duty Rosters	67.8±6.2	–	$26$0.54 run	160.69	T1764 · 8 Sep 2026prompt.txt
Fraktur Adverts	36.5±29.4	–	$23$0.12 run	150.33	T1754 · 8 Sep 2026prompt_optimized.txt
General Meeting Minutes	79.5±13.4	–	$14$0.13 run	89.62	T1763 · 8 Sep 2026prompt.txt
Library Cards	88.2±1.4	–	$3.35$0.88 run	18.65	T1755 · 8 Sep 2026prompt.txt
Magazine Pages	67.4±12.2	–	$4.36$0.20 run	28.13	T1762 · 8 Sep 2026prompt.txt
Medieval Manuscripts	55.0±13.5	–	$20$0.24 run	132.91	T1756 · 8 Sep 2026prompt.txt
Personnel Cards	87.2±6.1	–	$6.21$0.38 run	34.57	T1761 · 8 Sep 2026prompt.txt

Price

From	Input, $ per million tokens	Output, $ per million tokens	Source
8 Sep 2026	0.44	1.32	archived price page

Every run