Pricing

Stop Overpaying for AI

Multiple suppliers compete to serve your requests. You get the best price per token.

Your first 200 Million Tokens are on us*

Sign up and start building. Market pricing means you get more performance for every dollar you spend. Integrate and send your first request in <5 mins.

Get Started

*Free token count is based on $25 worth of sign up credits used for Text Prime tokens, count for other instruments may vary.

Text Standard

serving

DeepSeek V4 Flash
Qwen3.5 35B A3B
GPT-OSS 120B

Price-optimized, high-throughput text

Save up to 87%

$0.0351M Tokens²

Text Prime

serving

MiniMax-M3
Kimi K2.6
GLM-5.1
MiniMax-M2.7

Reliable models for everyday tasks

Save up to 78%

$0.1201M Tokens²

Text Max

serving

Claude Opus 4.8

Frontier models for complex reasoning

Save up to 16%

$1.7091M Tokens²

Code Standard

Preview

serving

DeepSeek V4 Flash
Qwen3.5 35B A3B
GPT-OSS 120B

Price-optimized, high-throughput code

Save up to 79%

$0.0351M Tokens²

Code Prime

Preview

serving

MiniMax-M3
Kimi K2.6
MiniMax-M2.7

Reliable models for daily coding work

Save up to 78%

$0.1201M Tokens²

Code Max

Preview

serving

Claude Opus 4.8

Frontier models for complex coding tasks

Save up to 50%

$1.7091M Tokens²

Agent Standard

Preview

serving

DeepSeek V4 Flash
Qwen3.5 35B A3B
GPT-OSS 120B

Price-optimized, high-throughput agents

Save up to 79%

$0.0351M Tokens²

Agent Prime

Preview

serving

MiniMax-M3
GLM-5.1
Kimi K2.6
MiniMax-M2.7

Reliable models for daily agent work

Save up to 82%

$0.1201M Tokens²

Agent Max

Preview

serving

Claude Opus 4.8

Frontier models for complex agent tasks

Save up to 16%

$1.7091M Tokens²

¹Compared to consuming similar-grade models directly from individual providers.

²Prices are refreshed frequently, but they move in real time with the market. For the latest price, visit The Grid App.

³The qualifying model list changes as new models meet the instrument spec and existing models are updated. Any model and supplier that qualifies the performance specification can serve requests on that instrument. The models listed here were recently active but are not guaranteed at any given time.