DeepSeek V4.1 Flash deepseek-v4.1-flash

deepseek-ai · deepseek

unranked — no quality signals yet

params
763.2B
arch
moe
context
1024k
license
mit open weights
released
Sep 2026
downloads/30d
2.1M

# architecture

attn shared GQA 64q/1kv router top-6 of 384 +1 shared expert ×384 expert out ×40 layers ctx 1,048,576 763.2B pool
layers
40
d_model
5120
heads
64q / 1kv
head dim
512
experts
top-6 of 384 + 1 shared
vocab
129280
family
deepseek_v41_text

signal path of one layer, generated from the registry's structured fields — dimension lines quote the model's real numbers; an MoE trace forks at the router, a dense trace runs straight through. Geometry fields come from the repo's config.json.

# trend

OGM score · last 51 days
OGM score over 51 days

# benchmarks

benchmarkscoresourcedate
Agents-last-exam 0.3 / 1 LLM Stats —
Automationbench 0.5 / 1 LLM Stats —
Babyvision 0.9 / 1 LLM Stats —
Chartography 0.8 / 1 LLM Stats —
Codeforces 3471.0 / 3000 LLM Stats —

# where to run

providerquantctx$/M in$/M out$/M cacheprice srctpsuptime✓
Featherless — ——— via hfrouter — —
Morphfp8 1024k $0.05$0.19$0.01 via openrouter — 100.0%
Sail Researchfp4 1024k $0.09$0.25$0.01 via openrouter — —
Sail Researchfp4 1024k $0.09$0.25$0.01 via openrouter — —
Sail 1024k $0.09$0.30— via litellm — —
OpenInferencefp4 1024k $0.04$0.50$0.01 via openrouter — 99.7%
Relacefp4 1024k $0.04$0.50$0.01 via openrouter — 100.0%
Wafermxfp4 1024k $0.04$0.50$0.01 via openrouter — 100.0%
Clf Ai Gateway 1024k $0.12$0.48— via litellm — —
InferenceNetfp8 1040k $0.10$0.60$0.01 via openrouter — 100.0%
InferenceNetfp8 1040k $0.10$0.60$0.01 via openrouter — 100.0%
StreamLakefp8 1000k $0.15$0.59$0.00 via openrouter — 100.0%
CoreWeavefp8 1024k $0.13$0.65$0.00 via openrouter — 100.0%
Scx 1000k $0.15$0.60— via litellm — —
Vultr 1024k $0.15$0.60— via modelsdev — —
Prism 1000k $0.17$0.63— via litellm — —
Alibabaunknown 1000k $0.17$0.66$0.02 via openrouter — 100.0%
Rekaunknown 1024k $0.10$0.84$0.01 via openrouter — 100.0%
DeepInfrafp8 1024k $0.20$0.60— deepinfra — 100.0% ✓
DeepInfrafp8 1024k $0.20$0.60— deepinfra — 100.0% ✓
DeepInfrafp8 1024k $0.20$0.60— deepinfra — 100.0% ✓
Wafermxfp4 1024k $0.19$0.65$0.01 via openrouter — 100.0%
Coralbricks 1024k $0.01$1.20— via litellm — —
Wandb 1049k $0.20$0.65— via litellm — —
GMICloudfp8 1048k $0.18$0.72$0.00 via openrouter — 97.8%
Fireworksunknown 1024k $0.21$0.84$0.00 via openrouter — 100.0%
Aiand 1024k $0.30$0.60— via litellm — —
DigitalOceanunknown 1024k $0.23$0.90$0.01 via openrouter — 100.0%
Mancer 2fp8 1024k $0.20$1.00— via openrouter — 98.4%
Kyma 1024k $0.27$0.81— via litellm — —
Makorafp8 1024k $0.26$1.17$0.01 via openrouter — 99.1%
Crusoefp8 1024k $0.29$1.20$0.01 via openrouter — 100.0%
AtlasCloudfp8 1024k $0.30$1.20— atlascloud — 100.0% ✓
Azure 1000k $0.30$1.20$0.01 via modelsdev — —
Baidufp8 1024k $0.30$1.20$0.01 via openrouter — 100.0%
BaseTenunknown 1024k $0.30$1.20$0.01 via openrouter — 99.8%
BaseTenunknown 1024k $0.30$1.20$0.01 via openrouter — 99.9%
Cohereunknown 1024k $0.30$1.20$0.00 via openrouter — 100.0%
DeepSeek 1000k $0.30$1.20$0.01 via requesty — —
Fireworksunknown 1024k $0.30$1.20$0.01 via openrouter — 97.9%
Googleunknown 1024k $0.30$1.20$0.03 via openrouter — 100.0%
Modalunknown 1024k $0.30$1.20$0.03 via openrouter — 100.0%
Nebiusfp8 1024k $0.30$1.20$0.01 openrouter — 97.9% ✓
Novitafp8 1024k $0.30$1.20— novita — 100.0% ✓
Opencode Go 1000k $0.30$1.20— via litellm — —
Parasailfp8 1024k $0.30$1.20$0.01 via openrouter — 100.0%
Phalaunknown 1024k $0.30$1.20— phala — 97.7% ✓
PrimeIntellectunknown 1024k $0.30$1.20$0.01 via openrouter — 100.0%
SiliconFlowfp8 1024k $0.30$1.20$0.01 openrouter — 99.9% ✓
Tencent 1000k $0.30$1.20— via litellm — —
Togetherunknown 1024k $0.30$1.20$0.01 via openrouter — 100.0%
Venicefp8 1000k $0.30$1.20— venice — 99.8% ✓
Zerogpu 1024k $0.30$1.20— via litellm — —
Inferenceapis 1024k $0.40$1.60— via litellm — —
Fireworksunknown 1024k $0.45$1.80$0.01 via openrouter — 100.0%

sorted by blended price ((3·input + output) / 4 per 1M) · ✓ = the provider's own catalog confirms the offer · "via …" prices are what the aggregator routing the offer charges, not the provider's own list price

source aliases
llmstats
deepseek-v4.1-flash
openrouter
deepseek/deepseek-v4.1-flash