Join us for our inaugural conference, Forge 2026

Pricing to seamlessly scale from idea to enterprise

Start building in seconds, self-serve. Contact us for enterprise deployments with faster speeds, lower costs, and higher rate limits.

Get started
Contact Us

Serverless Inference

Get started in seconds with per token pricing, zero setup and no cold starts

See Pricing

Training

Customize open models with your own data with minimal setup

See Pricing

On Demand Deployments

Pay per GPU second for faster speeds, higher rate limits, and lower costs at scale

See Pricing

Serverless Inference

Pay per token with pre-paid, usage-based billing. Simply add a payment method to buy credits, and usage is then deducted from your balance. Usage is billed on output, input, and cached tokens. Turn on Auto Reload to top up automatically when your balance runs low, or set a monthly spend limit to cap usage.

To view the current pricing for our most popular serverless models across Standard, Priority, and Fast serving paths, visit our full pricing documentation.

Embeddings

Base model parameter count$ / 1M input tokens

up to 150M

$0.008

150M - 350M

$0.016

Qwen3 8B

$0.1

Training Pricing

Serve fine-tuned models for the same price as base models.

Managed Training

Supervised and preference fine tuning is priced per 1M training tokens.

Base ModelLoRA SFTLoRA DPOFull Param SFTFull Param DPO

Models up to 16B parameters

$0.50

$1.00

$1.00

$2.00

Models 16.1B - 80B

$3.00

$6.00

$6.00

$12.00

Models 80B - 300B (e.g. Qwen3-235B, gpt-oss-120B)

$6.00

$12.00

$12.00

$24.00

Models >300B (e.g. DeepSeek V3, Kimi K2)

$10.00

$20.00

$20.00

$40.00

  • •SFT and DPO prices are shown in $ per 1M training tokens. Training tokens can be estimated with number of tokens in training dataset * number of epochs. Estimation should be multiplied by the average number conversation turns /2 for tuning with intermediate thinking traces.
  • •Please note that when fine-tuning with reasoning traces, including the reasoning_content field for assistant turns will increase the total number of tuned tokens because multi-turn conversations are unrolled into user, assistant, and thinking traces. For further details, please refer to example 2 in the documentation about SFT fine tuning.
  • •Fine-tuning with images (VLM supervised fine-tuning) is also billed per 1M tokens. See this FAQ on calculating image tokens.

Serverless Training API

Attach to a shared, always-on trainer pool for LoRA training on the launch models. There's no provisioning and no idle cost. You pay only for the tokens you prefill, sample, and train.

Base ModelContextPrefill / 1MCached Prefill / 1MSample / 1MTrain / 1M

GLM 5.3 Flash

200K

$2.96

$0.593

$7.41

$8.89

GLM 5.3

262K

$4.86

$0.972

$12.15

$14.58

Qwen 3.8 27B

128K

$1.86

$0.372

$5.595

$4.103

Kimi K3

192K

$10.87

$2.17

$27.11

$32.55

DeepSeek V4 Flash 0731

262K

$1.74

$0.35

$4.33

$5.20

Muse Glimmer 30B

128K

$1.96

$0.39

$4.88

$5.86

  • •Checkpoint storage for serverless models is included during private preview.
  • •Other frontier models are coming soon to the Serverless Training API catalog.

Dedicated Training API

Dedicated Training API jobs are priced per GPU hour. Please see the On-Demand Pricing section below for details on Dedicated Training API Pricing.


On-Demand Pricing

Pay per GPU second, with no extra charges for start-up times

On demand deployments

GPU TypePrice ($) per minutePrice ($) per hour

H100 80 GB GPU

$0.134

$8.00

H200 141 GB GPU

$0.134

$8.00

B200 180 GB GPU

$0.217

$13.00

B300 288 GB GPU

$0.250

$15.00

GB300 288 GB GPU

$0.334

$20.00

  • •Region-restricted deployments are priced at a 1.5x premium. Contact Sales to get started.