EDGE

Inference Pricing

OpenAI-compatible LLMs on owned GPUs. Per token, per model. Open-source models catching up with frontier at a fraction of the cost, with zero egress to storage and inference.

SIGN UP FOR FREE GET VOLUME RATE

ADJUST SETTINGS

Input tokens

01 T

Cached tokens

0100%

Output tokens

01 T

LLM

ESTIMATED COST

$368Per month

Input tokens150M × $0.665 / 1M: $99.75
Cached input tokens350M × $0.08 / 1M: $28.00
Output tokens60M × $4.00 / 1M: $240.00

SIGN UP FOR FREE GET VOLUME-BASED RATE

Published rates, pay as you go, before volume tiers. Caching means large prompts cost far less than list rate; actual billing is per token at the rates in the table below, with no prompt-size multipliers and no minimums.

Rates

What's on the card.

Each row is a line on your invoice. Nothing is bundled into another primitive's rate and nothing turns on unless you call it.

Chat completions pricing (per 1M tokens)

Chat completions pricing (per 1M tokens)

Model	Price
GLM-5.3 — Flagship frontier intelligence	Input: $1.400 / 1M tokens Cached Input: $0.260 / 1M tokens Output: $4.400 / 1M tokens
GLM-5.3 Flash — Fast, efficient inference	Input: $0.150 / 1M tokens Cached Input: $0.030 / 1M tokens Output: $0.500 / 1M tokens
DeepSeek V4.1 Flash — Low-latency multimodal inference	Input: $0.300 / 1M tokens Cached Input: $0.006 / 1M tokens Output: $1.200 / 1M tokens
Kimi K3 — Multimodal intelligence, 1M context	Input: $2.700 / 1M tokens Cached Input: $0.270 / 1M tokens Output: $13.500 / 1M tokens
Kimi K2.6 — Highest intelligence, voice AI	Input: $0.665 / 1M tokens Cached Input: $0.080 / 1M tokens Output: $4.000 / 1M tokens
GLM-5.2 — Next-gen efficient reasoning	Input: $1.000 / 1M tokens Cached Input: $0.200 / 1M tokens Output: $4.000 / 1M tokens
MiniMax-M3 — Cheapest while maintaining high intelligence	Input: $0.270 / 1M tokens Cached Input: $0.080 / 1M tokens Output: $1.100 / 1M tokens
DeepSeek V4 Flash — Ultra-fast, cost-efficient inference	Input: $0.130 / 1M tokens Cached Input: $0.030 / 1M tokens Output: $0.260 / 1M tokens
Qwen 3.8 27B — Lightweight open-weight reasoning	Input: $0.400 / 1M tokens Cached Input: $0.050 / 1M tokens Output: $3.000 / 1M tokens

Chat completions pricing (per 1M tokens)

Model	Price
GLM-5.2 — Next-gen efficient reasoning	Input: $2.000 / 1M tokens Cached Input: $0.400 / 1M tokens Output: $8.000 / 1M tokens

Other services

Service	Price
Embeddings (gte-large)	$0.0001 / 1K tokens
Speech to text	$0.003 / minute
AI-enabled storage and retrieval	$0.02 / GB / day

Machine readable: GET /v2/public/pricing. Rates may vary by destination and volume tier. Machine readable: GET /v2/public/pricing?primitive=inference.

VOLUME

Commit more, pay less per unit. Nothing is gated.

PAY AS YOU GO

$0

Standard rates. Card on file. Automatic tier discounts as usage crosses thresholds, no contract needed.

COMMITTED

from $500/ mo

Discounted rates across everything you use, not one product. Named account manager, priority support, higher limits.

GET A VOLUME RATE

ENTERPRISE

from $5,000/ mo

Dedicated infrastructure and IPs, private interconnect, in region GPUs, unlimited rate limits.

Usually bought with

What people add next.

See the whole composition on private ai deployment ↗

Live in five minutes. No sales gate in front of the API.

Card on file, provisioned by API. When you want a human, they run the network and pick up 24/7.

SIGN UP FOR FREE TALK TO SALES