Granite API pricing
Granite API pricing on 2 platforms: 16 Granite endpoints by IBM and 31 current prices. The cheapest exact price is $0.03/M in for granite-3.3-8b-instruct, on Replicate. Prices are read from each platform daily and tagged by task, resolution and duration.
- Endpoints
- 16
- Current prices
- 31
- Platforms
- 2
- Cheapest
- $0.03/M in
Granite API pricing by platform
One row per model, matched across platforms by its model id; each cell is input / output per million tokens. Green is the cheapest platform for that model. Different models are never ranked against each other.
| Model | OpenRouter | Replicate |
|---|---|---|
| Granite 4.2 8B | $0.06 in$0.25 out | $0.06 in$0.25 out |
Sold on one platform only (14)
Nothing to compare these with yet, so no price here is marked cheapest.
| Model | Platform | USD |
|---|---|---|
| granite-4.1-8b | Replicate | $0.06 in$0.25 out |
| ibm-granite/granite-speech-4.1-2b | Replicate | $0.03 in$0.25 out |
| granite-vision-4.1-4b | Replicate | $0.06 in$0.25 out |
| granite-4.0-h-small | Replicate | $0.06 in$0.25 out |
| granite-3.3-8b-instruct | Replicate | $0.03 in$0.25 out |
| ibm-granite/granite-speech-3.3-8b | Replicate | $4.9/M out |
| granite-vision-3.3-2b | Replicate | $0.25/M out |
| granite-3.2-8b-instruct | Replicate | $0.03 in$0.25 out |
| granite-3.1-2b-instruct | Replicate | $0.03 in$0.25 out |
| granite-3.1-8b-instruct | Replicate | $0.03 in$0.25 out |
| granite-3.0-2b-instruct | Replicate | $0.03 in$0.25 out |
| granite-3.0-8b-instruct | Replicate | $0.05 in$0.25 out |
| granite-20b-code-instruct-8k | Replicate | $0.1 in$0.5 out |
| granite-8b-code-instruct-128k | Replicate | $0.05 in$0.25 out |
Granite API pricing FAQ
How much does the Granite API cost?
Granite 4.2 8B: $0.06/M in ($0.25 output) on 2 platforms alike. Each price is the platform’s own list price, read daily.
Which platform has the cheapest Granite API?
OpenRouter is the cheapest in all 1 like-for-like comparisons.
Which platforms sell the Granite API?
2 of the platforms tracked here: Replicate, OpenRouter. Together they list 16 endpoints and 31 prices.
How current are these prices?
A collector reads each platform’s pricing API or page daily and keeps the response. Every price links to the listing it came from, so you can check it before you call.
Claude
Gemini
DeepSeek
Qwen
Llama
Mistral
Kimi
GLM
MiniMax
Hunyuan
Nemotron