impulse

Granite 4 H Small

Open weights
API id granite-4-h-small
Input / 1M
$0.0636
at IBM watsonx.ai
Output / 1M
$0.265
at IBM watsonx.ai
Context window
131K
tokens
Max output
131K
tokens per reply
Hosts
1
1 listings
Released
-
date not published

Pricing

List prices at IBM watsonx.ai, in US dollars per million tokens.

TierInput / 1MOutput / 1M
Standard$0.0636$0.265

What it costs a month

At IBM watsonx.ai list prices, 30 days.

  • Chat assistant5,000 requests a day, 2K in / 0.5K out$38.95
  • RAG over documents2,000 requests a day, 8K in / 0.6K out$40.07
  • Coding agent1,000 requests a day, 40K in / 2K out$92.22
  • Overnight batch20,000 requests a day, 4K in / 0.4K out$216
Use your own numbers

Where to run it

1 host serve Granite 4 H Small. The comparison is on blended price, 3 parts input to 1 part output.

HostIn / out per 1M
granite-4-h-small
$0.0636 / $0.265

Capabilities

Core

  • Streaming
  • Function calling
  • JSON mode
  • Structured outputs

Multimodal

  • Vision
  • Image input
  • Audio
  • Image generation
  • Video

Advanced

  • Reasoning
  • Tool calling
  • MCP
  • Prompt caching
  • Batch API
  • Embeddings
Takes in
Text
Produces
Text
Features
Parallel tool calls
Open weights
License: Apache-2.0
Runtimes: Ollama, llama.cpp, vLLM

Cheaper alternatives

Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.

Same model, cheaper host

No host sells it for less than IBM watsonx.ai.

Smaller IBM models

No cheaper IBM model keeps its key capabilities.

Other frontier labs

No cheaper match from another frontier lab.

Open-weight picks

  • Nemotron 3.5 Lightning
    NVIDIA · on Nebius AI Studio
    $0.06 / $0.24
    Has every key capability
    8% less
  • MiniMax M3.1 Flash
    MiniMax · on Refuseless
    $0.05 / $0.25
    Has every key capability
    12% less
  • Qwen3.5 9B
    Alibaba Qwen · on EmpirioLabs AI
    $0.09 / $0.13
    Has every key capability
    12% less
  • Ministral 3B 2512
    Mistral AI · on Mistral AI
    $0.10 / $0.10
    Has every key capability
    12% less

Model line

Newer and older versions, at each lab's own price.

  1. UndatedGranite 4 H SmallLatest$0.0636 / $0.265
  2. UndatedGranite 4.0 H Small$0.06 / $0.25

Call it

Using IBM watsonx.ai's API.

Python
from litellm import completion

response = completion(
    model="watsonx/granite-4-h-small",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Quick answers

How much does Granite 4 H Small cost?

On IBM watsonx.ai, Granite 4 H Small costs $0.0636 per million input tokens and $0.265 per million output tokens.

Which APIs offer Granite 4 H Small?

Granite 4 H Small runs on 1 API: IBM watsonx.ai.

What is Granite 4 H Small's context window?

131K tokens, with up to 131K tokens in one reply.

Is Granite 4 H Small being retired?

No retirement date is scheduled for it on any API we track.

Is there a newer version of Granite 4 H Small?

No. Granite 4 H Small is the newest model in its line.

What are cheaper alternatives to Granite 4 H Small?

Models that cost less and keep most of what it can do: Nemotron 3.5 Lightning on Nebius AI Studio (8% less). Savings are on a mix of 3 input tokens to 1 output token.

What is Granite 4 H Small's API model id?

Use granite-4-h-small with IBM watsonx.ai.

Sources

Raw catalog entry for granite-4-h-small on IBM watsonx.ai
{
  "max_input_tokens": 131072,
  "max_output_tokens": 131072,
  "input_cost_per_token": 6.36e-8,
  "output_cost_per_token": 2.65e-7,
  "litellm_provider": "watsonx",
  "mode": "chat",
  "supports_function_calling": true,
  "supports_parallel_function_calling": true,
  "supports_vision": false,
  "source": "https://dataplatform.cloud.ibm.com/docs/content/wsj/analyze-data/fm-models.html?context=wx"
}