DeepSeek V4 Flash 0731
DeepSeek · Open weights · Released Jul 31, 2026 · deepseek/deepseek-v4-flash-0731
Best $0.01 in / $0.132 out via Relace · fp4
1.05M contextToolsReasoningStructured outputPrompt cachingOpen weights
Available from 28 providersUSD / MTOK · PRICES 23 MINUTES AGO · SPEED 3 MINUTES AGO
EST. MONTHLY COSTclick a row for detail · click headers to sort
| PROVIDER | INPUT | OUTPUT | CACHE | CONTEXT | QUANT | TTFT | TPS | SMOKE | UPTIME |
|---|---|---|---|---|---|---|---|---|---|
 Relace | $0.01 | $1.28 | $0.01 | 1.05M | fp4 | 0.69s° | 102° tok/s | 5/5 | 100.0% |
| OPOpenInference | $0.011 | $1.01 | $0.011 | 1.05M | fp4 | 5.3s° | 15° tok/s | 3/5 | 99.7% |
| SRSail Research | $0.019 | $0.42 | $0.012 | 1.05M | fp4 | 1.6s° | 50° tok/s | 5/5 | 93.8% |
 Reka | $0.021 | $0.528 | $0.006 | 262K | — | 0.81s° | 112° tok/s | — | 100.0% |
 StreamLake | $0.044 | $0.132 | $0.001 | 1.02M | fp8 | 2.8s° | 34° tok/s | 4/5 | 100.0% |
 Inceptron | $0.05 | $0.65 | $0.027 | 1.05M | fp4 | 1.3s° | 73° tok/s | — | 100.0% |
 DeepInfra | $0.06 | $0.18 | $0.015 | 1.05M | fp8 | 0.91s° | 47° tok/s | — | 99.6% |
| SRSail Research | $0.10 | $0.30 | $0.03 | 1.05M | fp4 | — | — | — | 93.1% |
 DigitalOcean | $0.119 | $0.238 | $0.024 | 1.05M | — | ✕ | ✕ | — | 99.7% |
 BaseTen | $0.13 | $0.26 | $0.028 | 1.05M | fp8 | 0.38s° | 265° tok/s | — | 99.9% |
| BaseTen | $0.13 | $0.26 | $0.028 | 1.05M | fp8 | — | — | — | 100.0% |
 CheaperInferenceDIRECTMARKETPLACE | $0.13 | $0.40 | — | — | — | 3.3s | 78 tok/s | — | — |
 CoreWeave | $0.13 | $0.28 | $0.07 | 262K | fp8 | ✕ | ✕ | — | 100.0% |
 Venice | $0.13 | $0.26 | $0.028 | 1M | — | ✕ | ✕ | — | 99.9% |
 WaferPRIORITY | $0.13 | $0.23 | $0.12 | 1.05M | — | 0.74s° | 100° tok/s | — | 100.0% |
 Cohere | $0.14 | $0.28 | $0.07 | 1.05M | — | ✕ | ✕ | — | 99.6% |
 Parasail | $0.14 | $0.28 | $0.05 | 1.05M | fp8 | ✕ | ✕ | — | 100.0% |
 Together | $0.14 | $0.28 | $0.03 | 1.05M | — | 18.3s° | 26° tok/s | — | 100.0% |
| CheaperInferenceDIRECTMARKETPLACE | $0.17 | $0.49 | — | — | — | 2.9s | 71 tok/s | — | — |
 Mancer 2 | $0.20 | $0.60 | — | 1.05M | fp8 | 0.78s° | 61° tok/s | — | 99.5% |
 SiliconFlow | $0.22 | $0.66 | $0.028 | 1.05M | fp8 | 1.8s° | 36° tok/s | — | 99.7% |
 GMICloud | $0.286 | $0.858 | $0.009 | 1.05M | fp8 | 3.1s° | 98° tok/s | — | 100.0% |
 Alibaba | $0.352 | $1.06 | $0.035 | 1M | — | ✕ | ✕ | — | 100.0% |
 Novita | $0.409 | $1.23 | $0.026 | 1.05M | fp8 | ✕ | ✕ | — | 100.0% |
 AtlasCloud | $0.44 | $1.32 | $0.028 | 1.05M | fp4 | 1.6s° | 237° tok/s | — | 99.8% |
 Baidu | $0.44 | $1.32 | $0.014 | 1.05M | fp8 | ✕ | ✕ | 4/5 | 100.0% |
 Cloudflare | $0.44 | $1.32 | $0.014 | 1.05M | — | 0.68s° | 86° tok/s | — | 100.0% |
 Phala | $0.44 | $1.32 | $0.028 | 1.05M | — | ✕ | ✕ | — | 99.9% |
TTFT/TPS: median (P50) over trailing 504h · chat10k profile (~10K in / 200 out) · measured against the provider API directly; ° = routed via OpenRouter (adds a hop) · ✕ = probed, endpoint returned no measurable stream · SMOKE: of 5 programmatic capability probes passed (JSON schema, instruction following, tool call, long-context, benign compliance; retry-once) · screened via OpenRouter routing · v0.2 · methodology v1.1 · measured 3 minutes ago
SERVE THIS MODEL AND NOT LISTED? LIST YOUR INFERENCE →
Quality benchmarksCURATED AUG 24, 2026
Third-party reported results, hand-curated with sources — not measured by ModelIndex. Our own measured serving data (TTFT/TPS/SMOKE) is in the providers table above.
Input price history · best listed price, observed
↓ 84% SINCE AUG 23, 2026
$0.064 · AUG 23, 2026$0.01 · TODAY
Pricing variants · USD / MTok in / out
STANDARD
$0.01 / $1.28
per MTok in / out
CACHE READ $0.01
Recent changesRSS ↗
Oct 8, 2026price decreased · input0.018 → 0.0104
Oct 7, 2026endpoint removed · morph31 → 28
Oct 7, 2026endpoint removed · nebius31 → 28
Oct 7, 2026endpoint removed · nextbit31 → 28
Oct 6, 2026endpoint added · nebius30 → 31
Oct 3, 2026endpoint removed · makora31 → 30
Oct 2, 2026price decreased · input0.0045 → 0.0038
Oct 1, 2026price decreased · input0.01 → 0.0045
About
DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....
Specifications
Context window1,048,576
Max output943,718
Modalities intext
Modalities outtext
LicenseOpen weights
ReleasedJul 31, 2026
TokenizerDeepSeek
WeightsHugging Face ↗
Open weights
safetensors
Related
Compare with
Embed badge
[](https://modelindex.ai/models/deepseek/deepseek-v4-flash-0731)Sources
OpenRouter API23 minutes ago
AGGREGATORHugging Face23 minutes ago
COMMUNITYModelIndex benchmarks14 days ago
MEASUREDVerified 23 minutes ago · list prices, not negotiated rates.