Granite 4.0 H Small
Granite 4.0 H Small is IBM's 32B-parameter hybrid Mamba/transformer mixture-of-experts instruct model, designed to respond to general instructions and to build AI assistants for multiple domains including business applications.
text → text · made by IBM watsonx.ai
Sold by 2 ways
| Seller | Lane | Rate |
|---|---|---|
| IBM watsonx cloud | standard | $0.064per Mtok in$0.27per Mtok outmodels.dev · read 2026-09-16 |
| IBM watsonx.ai cloud | standard | $0.064per Mtok in$0.27per Mtok outibm.com · read 2026-08-25 |
Measured 2 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 13thof 105 | Vectara Hallucination Leaderboard | Hallucination Rate | 5.2 |
| 264thof 302 | Artificial Analysis · Intelligence Index | Artificial Analysis Intelligence Index | 6 |
About
Granite 4.0 H Small — a text model from IBM watsonx.ai, sold by 2 companies from $0.064 in and $0.27 out per million tokens, placed 13th of 105 on Vectara Hallucination Leaderboard.
It takes text and returns text, with a context window of 128,000 tokens. It was published in October 2025. Its sellers say it can call a tool. The catalogue files it under chat. Two companies sell it. The cheapest is $0.064 in and $0.27 out per million tokens at IBM watsonx. It has been measured on 2 boards, and stands best at 13th of 105 on Vectara Hallucination Leaderboard.
Every current figure
- Maker
- IBM watsonx.ai
- Register
- model
- Takes
- text
- Returns
- text
- Context
- 128,000 tokens
- Published
- October 2025
- Licence
- not read
- Sellers
- 2
- Price
- $0.064 in and $0.27 out per million tokens — IBM watsonx
- Boards
- 2
- Best place
- 13th of 105 — Vectara Hallucination Leaderboard
Known as 1 name
granite-4h-small