NewForge and Unlimited plans: more from every dollar, on models already below market price.Forge and Unlimited plans: more from every dollar.See plans

// BACKED BY

Y COMBINATOR //

Run Open-source models.At lowest cost. Highest speed.

Open models the way they should always have been: ultra-fast and too cheap to meter.

////////////////////

@@**__**__**@@

\\\\\\\\\\\\\\\\\\\\

Kimi
Z.ai
DeepSeek
MiMo
Qwen
MiniMax

WHY OPENSCALE

The most efficient place to run open models

Cheaper, faster, and fully elastic, so you ship on open weights without managing a single GPU

COST

TYPICAL

OPENSCALE

=== === ===

Lowest cost

Up to 50% cheaper than typical providers. We serve open weights efficiently and pass the savings straight to you.

===

PAYMENTS

===

1M tokens

$0.39

10M tokens

$3.90

100M tokens

$39.00

1B tokens

$390.00

linear pricing, billed only for tokens used

Pay as you go

Only pay for the tokens you use. No idle GPUs, no seats, no minimums, no commitments.

<<<<<<<

SPEED

>>>>>>>

TYPICAL

OPENSCALE200+ TOK/S

=== === ===

Fastest inference

200+ tokens per second on DeepSeek V4.1 Flash, with tuned kernels and warm capacity for low latency.

AUTOSCALING

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

^

Scale to infinity

Autoscaling that handles a single request or a million, with no capacity planning on your side.

MODELS & PRICING

Below market price, on every model.

The struck price is what the same tokens cost at market price. Same price around the clock.

DeepSeek V4.1 Flash

DeepSeek

IN / 1M
$0.06
market $0.15
CACHED / 1M
$0.003
market $0.003
OUT / 1M
$0.24
market $0.60
TOK / S
200+
2.5x cheaperVS MARKET PRICE

GLM 5.3 Flash

Z.ai

IN / 1M
$0.12
market $0.15
CACHED / 1M
$0.024
market $0.03
OUT / 1M
$0.40
market $0.50
TOK / S
100+
1.25x cheaperVS MARKET PRICE

Qwen3.8 27B

Alibaba

IN / 1M
$0.04
market $0.21
CACHED / 1M
$0.01
market $0.04
OUT / 1M
$1.25
market $2.55
5x cheaperVS MARKET PRICE

Kimi K3.1 Pro

Moonshot AI

IN / 1M
$0.30
market $2.00
CACHED / 1M
$0.20
market $0.20
OUT / 1M
$9.00
market $10.00
6x cheaperVS MARKET PRICE

SAVINGS

What you save each month

Set your monthly tokens. See what you spend, what you save, and how a plan saves more.

Your usage
Model

Both sides at published list prices.

You spend $981 a month. That’s $1,350 less than market price. On Forge 500, you’d save $200 more.

Suits your usage

Forge 500Instead of $981$781/mo

  • Covers $700 of your $981 month. You pay $500 for the plan plus the other $281.
  • Every model, standard or fast mode.
  • 20% less than pay-as-you-go for this month.
Start for $100 (Forge 500 plan)Or stay on pay-as-you-go at $981

Running your own agents all day? Unlimited is one flat price, from $39 a month. See Unlimited

OPENSCALE PLANS

Token Plans
Cheaper, faster, unlimited

Forge adds up to 50% more usage to what you pay. Unlimited runs your own agents all day on one flat price.
Fastest speedUp to 3x market price valueForgefrom$25/moPower your AI apps and agents with the fastest speedYou are building an AI app or productYour agents need answers fastUp to 50% more usage than you pay for, on every modelSee Forge plansAlways workingUnlimited tokensUnlimitedfrom$39/moThe most work done, for agents that are always workingYour own agents run all day, every dayOne flat price, no usage to countNot included: Personal use only: not for apps, reselling or shared keysSee Unlimited plans

HOW IT WORKS

From zero to inference in minutes

If you can call the OpenAI API, you can run on OpenScale. Three steps, no infrastructure to manage.

Pick a model

Choose from every major open model, all behind one OpenAI-compatible endpoint.

Point your code

Swap your base URL and key. Keep your existing OpenAI SDK, prompts and tooling exactly as they are.

Scale automatically

We handle batching, warm capacity and autoscaling. You ship, from the first request to millions.

ZERO DATA RETENTION

Your prompts stay yours

Chat and Messages requests are processed in memory and never stored. Responses API responses are kept 30 days for multi-turn state, and one switch turns that off.

  • 01Nothing stored

    Chat Completions and Messages bodies are never written to disk. Only 30 days of metadata is kept, for billing.

  • 02Gone after the reply

    A chat request lives only as long as it takes to answer it. Once the tokens are counted, the content is discarded.

  • 03Every model, one switch

    Every model runs with no retention where it is served. Zero data retention turns Responses API storage off too. Your data never trains anything.

SANDBOXES

Built-in with more features.
At 20x cheaper price.

Compared with the sandbox providers below.

100 agents× 1 hour

4 vCPU · 8 GB each · 100 at once by default

Provider100 × 1 hMultiple of OpenScale
OpenScale$3.40
E2B$33.10×9.7
Daytona$33.10×9.7
Modal Sandboxes$47.60×14
Runloop$63.40×18.6
Vercel Sandbox$68.20×20.1

List prices, 4 vCPU · 8 GB, 3 Nov 2026. Vercel at full CPU.

Included on every sandbox

No caps

  • No session time limit
  • 100 at once from day one
  • 25 GB disk included

Resumable

  • Sleeps when idle, wakes on use
  • Resumes with memory and files intact
  • Paused sandboxes are free

No overhead

  • No platform fee, billed per second
  • Same key as your models
  • Your SDK code runs unchanged

Also on every sandbox: commands, streaming, an interactive terminal, files, preview URLs, Dockerfile templates, a headless browser, pause and resume.

// OpenScale for Startups //

Apply for OpenScale
Startup Program

Dear founders,

OpenScale gives early stage startups up to $5,000 in free credits to build things.

Run open models for your product and coding agents, at the lowest price anywhere. Apply to the program and you get:

Non Y Combinator startup

  • $25 to start
  • First top-up matched 1:1 up to $250
  • 20% back every month for 6 months
  • Up to $5,000 in free credits

In Y Combinator

  • $50 to start
  • First top-up matched 1:1 up to $500
  • 20% back every month for 6 months
  • Up to $5,000 in free credits

PROGRAMS

Build on us, or resell us.

Extra credit for the teams that bring us the most work.
Startup Program

Up to $5,000 in free credits

For early stage startups on open models: $25 to start, your first top-up matched up to $250, then 20% back every month.

Reseller Program

Up to 60% extra credit on every top-up

For routers, relays, gateways and aggregators: 20% to 60% extra credit by volume, higher limits from day one and our written OK to resell.