ZMIME
Comparison · 3 models · Updated Oct 4, 2026

Granite-4.0-H-Small vs Llama 3.3 Nemotron Super 49B v1.5 vs Apertus 8B

Granite-4.0-H-Small comes out ahead, 63 to 56 and 50 on our weighted score, and it is the cheaper option too.

  1. Our pick

    IBM

    Granite-4.0-H-Small

    Released Oct 2, 2025

    63/100
    • ECI—
    • Price$0.064 / $0.265
    • Context131K
    Full specs Remove Granite-4.0-H-Small

  2. NVIDIA

    Llama 3.3 Nemotron Super 49B v1.5

    Released Jul 25, 2025Deprecated

    50/100
    • ECI—
    • Price$0.40 / $0.40
    • Context131K
    Full specs Remove Llama 3.3 Nemotron Super 49B v1.5

  3. Swiss AI

    Apertus 8B

    Released Sep 2, 2025

    56/100
    • ECI—
    • Price$0.10 / $0.20
    • Context66K
    Full specs Remove Apertus 8B
01 — Verdict

Granite-4.0-H-Small is our pick

Granite-4.0-H-Small is the better all-round choice, scoring 63/100 against Apertus 8B (56) and Llama 3.3 Nemotron Super 49B v1.5 (50). It leads on price. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.

  • CapabilityNot enough dataNo independent benchmark covers every model here yet
  • Lowest priceGranite-4.0-H-SmallGranite-4.0-H-Small $0.114 · Apertus 8B $0.125 · Llama 3.3 Nemotron Super 49B v1.5 $0.40 per 1M tokens (3:1 blend)
  • Longest contextGranite-4.0-H-Small and Llama 3.3 Nemotron Super 49B v1.5Granite-4.0-H-Small 131,072 · Llama 3.3 Nemotron Super 49B v1.5 131,072 · Apertus 8B 65,536 tokens
  • Widest inputsSame inputsGranite-4.0-H-Small: Text · Llama 3.3 Nemotron Super 49B v1.5: Text · Apertus 8B: Text
  • Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
How the score is built
MeasureWeightGranite-4.0-H-SmallLlama 3.3 Nemotron Super 49B v1.5Apertus 8B
Price50%956993
Inputs & features30%353525
Context window20%242412
Overall100%63/10050/10056/100

Left out because at least one model lacks the data: capability. The remaining weights were rescaled.

02 — Side by side

Every spec in one table

Highlighted cells lead their row. Dashes mean the data is not published.

Granite-4.0-H-Small vs Llama 3.3 Nemotron Super 49B v1.5 vs Apertus 8B specifications side by side
SpecificationGranite-4.0-H-SmallIBMLlama 3.3 Nemotron Super 49B v1.5NVIDIAApertus 8BSwiss AI
Capability
Capabilities Index (ECI)———
ECI rank———
Price per million tokens
Input$0.064 (best)$0.40$0.10
Output$0.265$0.40$0.20 (best)
Cached input———
Blended (3:1)$0.114 (best)$0.40$0.125
Long-context rateSame rateSame rateSame rate
Price sourceOfficial watsonx.ai APIMedian of 1 providersMedian of 1 providers
Limits
Context window131,072 tokens (best)131,072 tokens (best)65,536 tokens
Max output131,072 tokens (best)131,072 tokens (best)8,192 tokens
Inputs and features
TextYesYesYes
ImagesNoNoNo
PDFsNoNoNo
AudioNoNoNo
VideoNoNoNo
ReasoningNoYesNo
Tool callingYesYesYes
Structured outputYesNoNo
Availability
WeightsOpenOpenOpenApache-2.0
API model IDibm/granite-4-h-smallnvidia/llama-3.3-nemotron-super-49b-v1.5—
API providers12 (best)1
ReleasedOct 2, 2025Jul 25, 2025Sep 2, 2025
Knowledge cutoff——Sep 2025
03 — Cost

What would a month cost?

Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.

  • Granite-4.0-H-Small$1.17
  • Llama 3.3 Nemotron Super 49B v1.5$4.80
  • Apertus 8B$1.40
04 — Questions

Which should you choose?

Which is better: Granite-4.0-H-Small, Llama 3.3 Nemotron Super 49B v1.5 or Apertus 8B?

Granite-4.0-H-Small is the better all-round choice, scoring 63/100 against Apertus 8B (56) and Llama 3.3 Nemotron Super 49B v1.5 (50). It leads on price. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.

Which is cheaper, Granite-4.0-H-Small, Llama 3.3 Nemotron Super 49B v1.5 or Apertus 8B?

Granite-4.0-H-Small is cheaper at $0.064 input / $0.265 output per million tokens (official watsonx.ai API price). Apertus 8B costs $0.10 input / $0.20 output per million tokens (median across 1 API provider); Llama 3.3 Nemotron Super 49B v1.5 costs $0.40 input / $0.40 output per million tokens (median across 1 API provider; free on Nvidia). At a typical mix of three input tokens to one output token, that is $0.114 per million tokens for Granite-4.0-H-Small versus $0.125 for Apertus 8B (1.1× as much) and $0.40 for Llama 3.3 Nemotron Super 49B v1.5 (3.5× as much).

Which scores higher on benchmarks?

There is no independent benchmark that covers all three models yet. Granite-4.0-H-Small has not been scored yet, Llama 3.3 Nemotron Super 49B v1.5 has not been scored yet and Apertus 8B has not been scored yet.

Which is better for coding?

There are no published SWE-bench Verified results for Granite-4.0-H-Small, Llama 3.3 Nemotron Super 49B v1.5 and Apertus 8B yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. All three support tool calling for agent workflows.

Which has the bigger context window?

Granite-4.0-H-Small and Llama 3.3 Nemotron Super 49B v1.5 have the largest context windows (131,072 and 131,072 tokens), against 65,536 for Apertus 8B. Maximum output per response: Granite-4.0-H-Small up to 131,072, Llama 3.3 Nemotron Super 49B v1.5 up to 131,072, Apertus 8B up to 8,192 tokens.

Which can read images, PDFs, audio or video?

Granite-4.0-H-Small accepts text; Llama 3.3 Nemotron Super 49B v1.5 accepts text; Apertus 8B accepts text. They handle the same number of input types.

Are any of these open source?

Yes, all three publish their weights (Apache-2.0), so you can self-host them.

Which is newer?

Granite-4.0-H-Small is the newest, released Oct 2, 2025. Apertus 8B came out Sep 2, 2025; Llama 3.3 Nemotron Super 49B v1.5 came out Jul 25, 2025. Knowledge cutoff: Apertus 8B Sep 2025.

How do you decide the winner?

Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.

05 — Keep comparing

Related comparisons