DeepSeek V3.1 pricing.

prices as of · 12 providers · 12 prices

DeepSeek V3.1 holds 164k tokens of context and has a reasoning mode. Open weights, so anyone may host it, which is why the prices below differ so much.

DeepSeek V3.1 costs $0.25 in and $0.95 out per million tokens from OpenRouter; the cheapest price we can explain is $1.40 at TensorX, checked 16 Sept 2026.

tool callingreasoningstructured outputopen weights164k contextreleased 21 Aug 2025

OpenRouter list price

$1.70/M tokens

$0.25 in · $0.95 out

read · 52% below the market median

Cheapest price we can explain

$1.40/M tokens

$0.20 in · $0.80 out · 18% under list

TensorX

read

What it costs for your work

pages a month

OpenRouter list price

$0.207a month

cheapest we can explain · TensorX

$0.168a month

difference

$0.039saved

A page is about 600 tokens in and 60 out, a support reply 200 in and 300 out, an email 150 in and 250 out. Nothing else is counted: no cache discount, no batch rate, no free tier.

Every provider selling DeepSeek V3.1

Four groups, decided by the price and by what the provider publishes about it. Prices are per million tokens, and the share is that row against OpenRouter's price. DeepSeek trained DeepSeek V3.1 and sells it on no catalog we read, so the price every row is measured against is OpenRouter's.

Cheaper, and we know why

1 price

Below list, with a published reason: hosting the open weights, a regional price list, credits, a stated discount.

  • TensorX

    Hosts the open weights on its own servers, so the price is its own, not a resale. Precision not disclosed.

    164k context here

    $0.20 / $0.80in / out per M

    0.82×of list price

Same price as the maker

10 prices

The same price the maker charges, sold by someone else.

  • OpenRouter

    The same price OpenRouter charges.

    $0.25 / $0.95in / out per M

    1.00×of list price

  • DeepInfra

    The same price OpenRouter charges.

    fp4 precisionDeepInfra

    $0.25 / $0.95in / out per M

    1.00×of list price

  • Nous Portal

    The same price OpenRouter charges.

    $0.25 / $0.95in / out per M

    1.00×of list price

  • Kilo Gateway

    8% above OpenRouter's price.

    $0.27 / $1in / out per M

    1.06×of list price

  • SiliconFlow

    8% above OpenRouter's price.

    fp8 precisionSiliconFlow

    $0.27 / $1in / out per M

    1.06×of list price

  • Novita

    8% above OpenRouter's price.

    131k context herefp8 precisionNovita

    $0.27 / $1in / out per M

    1.06×of list price

  • Atlas Cloud

    20% above OpenRouter's price.

    131k context herefp8 precisionAtlas Cloud

    $0.30 / $0.95in / out per M

    1.09×of list price

  • CoreWeave

    120% above OpenRouter's price.

    161k context herefp8 precision

    $0.55 / $1.65in / out per M

    1.94×of list price

  • Mara

    140% above OpenRouter's price.

    131k context here

    $0.60 / $1.70in / out per M

    2.06×of list price

  • Google

    140% above OpenRouter's price.

    $0.60 / $1.70in / out per M

    2.06×of list price

Cheaper, and nobody says why

1 price · never ranked

Below list on a closed model with nothing to account for it. Shown, never ranked, never a pick.

  • SambaNova

    We could not read this provider's prices reliably, so the row is listed but never ranked.

    131k context herefp8 precisionSambaNova

    $0.65 / $1.50in / out per M

    2.03×of list price

Frequently asked

Who sells DeepSeek V3.1 cheapest?
TensorX, at $0.20 in and $0.80 out per million tokens, read on 16 Sept 2026. Hosts the open weights on its own servers, so the price is its own, not a resale. Precision not disclosed.
Is DeepSeek V3.1 open weights?
Yes. DeepSeek publishes the weights, so anyone may host DeepSeek V3.1, which is why 12 providers sell it and why their prices differ so much.
How much context does DeepSeek V3.1 hold?
163,840 tokens in one call, which is about 164k, and it can call your own functions.
How does DeepSeek V3.1 compare with the rest of the market?
At $1.70 per million tokens, OpenRouter's price is 52% below the $3.55 median across the 337 text models we track.
More from DeepSeekDeepSeek V4 Flash 0731DeepSeek V4 ProDeepSeek V3.2DeepSeek V4.1 FlashSimilar contextgpt-oss-120bGLM-5.1DeepSeek V3.2GLM-5