DeepSeek language model

DeepSeek V4 Flash

retired 1 November 2026

live models in this family: DeepSeek V4.1 Flash, DeepSeek V4 Pro.

the smaller DeepSeek V4 model: an open-weights mixture-of-experts with 284B total and 13B active parameters that takes text only. on its own api, DeepSeek has replaced it with V4.1 Flash, which costs less and adds image input.

01

historical price

$0.44/M

input tokens

$1.32/M

output tokens

$0.014/M

cached input

historical listed price per 1M tokens.

  • the listed rate is DeepSeek's peak-hour price from its 2026-08-16 revision. it applies at all hours here, although DeepSeek's own off-peak rate was half.
  • DeepSeek retired the model on its own api on 2026-09-10, and that route now serves V4.1 Flash. this page is for the V4 Flash open weights served by third-party hosts.

02

specs

provider
DeepSeek
context window
1M tokens — ≈ 1,500 pages of prose or 100,000 lines of code
input
text
output
text
structured output
yes — JSON schema
reasoning mode
yes
prompt caching
yes — cache hits at $0.014/M
weights
open — MIT
released
April 24, 2026
status
retired 1 November 2026
model id
deepseek-v4-flash

integration notes

  • the weights ship without a Jinja chat template; prompts must be encoded with DeepSeek's encoding scripts.
  • Think Max needs a special system prompt and, per DeepSeek, a context window of at least 384K tokens.
  • DeepSeek recommends temperature 1.0 and top_p 1.0 for local deployment.

03

benchmarks

as published by the maker; not measured by octomind.

benchmarkscore
SWE Verified (Think Max)source79.0
Terminal Bench 2.0 (Think Max)source56.9
GPQA Diamond (Think Max)source88.1
HLE (Think Max)source34.8
BrowseComp (Think Max)source73.2

04

what it was for

pick it if

  • you want MIT-licensed open weights from DeepSeek's cheaper V4 tier
  • you need a 1M-token context with cached input at $0.014 per 1M
  • you want a choice of non-think, Think High or Think Max reasoning

05

compare DeepSeek V4 Flash

all model comparisons →

06

similar models

nearest live models by listed input price.

view all models →

07

DeepSeek V4 Flash faq

how much does DeepSeek V4 Flash cost?

DeepSeek V4 Flash listed at $0.44/M input, $1.32/M output, $0.014/M cached input per 1M tokens before it retired on 1 November 2026.

what is the context window of DeepSeek V4 Flash?

1M tokens, ≈ 1,500 pages of prose or 100,000 lines of code.

does DeepSeek V4 Flash accept images?

no. DeepSeek V4 Flash is text-only; pick a vision model from the DeepSeek lineup or the catalog for screenshots and diagrams.

is DeepSeek V4 Flash open source?

its weights are open under MIT, so it can also be self-hosted.

when was DeepSeek V4 Flash released?

April 24, 2026, by DeepSeek.

is DeepSeek V4 Flash still available?

no. it retired on 1 November 2026. current models in the family: DeepSeek V4.1 Flash and DeepSeek V4 Pro.

use DeepSeek V4 Flash in octomind

explore the live models in this family above to choose a current replacement.