Models

DeepSeek-V3

hyperbolic · 164K ctx · 12 providers · updated 2026-08-05 00:15 UTC

Cheapest
$0.200 Official list · Hyperbolic
Avg price
$0.841 /1M
providers
12
(3 × $0.200 + $0.200) ÷ 4 = $0.200/1M — the $/1M figure blends input and output at 3:1 — that is the only way models with two different rates can be ranked on one axis. Your actual rate depends on your own input:output mix — input only $0.200, 3:1 $0.200, 1:1 $0.200, output only $0.200 per 1M.

The board shows this model's cheapest measured offer ($0.200/1M); the smaller figure under it is the mean of the 14 offers listed below ($0.841/1M). Add those up and divide — it matches.

ProviderRouter / Source$ in /1M$ out /1MTTFTTPSUptimeEvidence
hyperbolic Pricebook (LiteLLM) $0.200$0.200 Official list
deepinfra HuggingFace Router $0.320$0.890 1021ms 20 API-measured
deepseek Pricebook (LiteLLM) $0.270$1.10 Official list
burncloud burncloud $0.287$1.15 API-measured
deepinfra Pricebook (LiteLLM) $0.380$0.890 Official list
novita HuggingFace Router $0.400$1.30 1171ms 33 API-measured
nebius Pricebook (LiteLLM) $0.500$1.50 Official list
requesty Requesty $0.850$0.900 API-measured
novita-ai Novita AI $0.890$0.890 API-measured
fireworks_ai Pricebook (LiteLLM) $0.900$0.900 Official list
vercel_ai_gateway Pricebook (LiteLLM) $0.900$0.900 Official list
together_ai Pricebook (LiteLLM) $1.25$1.25 Official list
replicate Pricebook (LiteLLM) $1.45$1.45 Official list
azure_ai Pricebook (LiteLLM) $1.14$4.56 Official list

Price trend

$0.2 $0.2 2026-07-06 → 2026-08-05 · 31d 2026-07-06 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-07 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-08 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-09 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-10 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-11 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-12 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-13 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-14 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-15 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-16 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-17 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-18 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-19 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-20 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-21 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-22 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-23 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-24 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-25 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-26 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-27 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-28 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-29 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-30 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-07-31 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-08-01 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-08-02 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-08-03 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-08-04 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M2026-08-05 · O 0.2 H 0.2 L 0.2 C 0.2 $/1M

Cheapest blended price, last 31 days (USD/1M): $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2 → $0.2

Data & method

Cheapest ranked by blended = (3×in + out) ÷ 4; performance columns show only when a source provides them, never estimated. “Official list” = LiteLLM open pricebook (MIT); “API-measured” = provider/aggregator public APIs. Machine-readable data: /api/v1/models/deepseek-v3.json. Free reuse requires attribution to tkx.org.

About

DeepSeek-V3 is an AI model designed for machine learning tasks, served by multiple inference providers on AI model marketplaces and listed under the "hyperbolic" serving-platform tag.

AI-generated summary from public information — is this yours?

FAQ

How much does DeepSeek-V3 cost?

Cheapest measured offer right now: $0.2/1M input, $0.2/1M output via hyperbolic — across 12 tracked providers, refreshed hourly.

Who serves DeepSeek-V3?

12 providers currently serve it. The table above lists each with input/output price, TTFT, TPS, uptime and an evidence tier.

Is the price of DeepSeek-V3 going up or down?

The daily candles above track the cheapest blended price ($/1M, 3:1 in:out). TKX snapshots every hour, so drops and hikes show up the same day they happen.

Related analysis

Back to rankings