ZMIME
Comparison · 3 models · Updated Nov 18, 2026

Hermes 3 70B Instruct vs Llama 3.1 Nemotron Ultra 253B vs Aya Expanse 32B

No independent capability benchmark covers every model here yet. The score uses price, inputs & features and context window.

Price in rating: Included · Exclude price

  1. Nous Research

    Hermes 3 70B Instruct

    Released Aug 18, 2024

    34/100
    • ECI—
    • Price$0.70 / $0.70
    • Context131K
  2. NVIDIA

    Llama 3.1 Nemotron Ultra 253B

    Released Apr 7, 2025Deprecated

    67/100
    • ECI—
    • PriceFree / Free
    • Context128K
  3. Cohere

    Aya Expanse 32B

    Released Oct 24, 2024

    33/100
    • ECI—
    • Price$0.50 / $1.50
    • Context128K
01 — Verdict

Capability evidence is incomplete

There is not enough shared capability evidence to name an overall winner. The scores (Llama 3.1 Nemotron Ultra 253B 67/100, Hermes 3 70B Instruct 34/100, Aya Expanse 32B 33/100) compare the available specifications and 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 priceLlama 3.1 Nemotron Ultra 253BLlama 3.1 Nemotron Ultra 253B Free · Hermes 3 70B Instruct $0.70 · Aya Expanse 32B $0.75 per 1M tokens (3:1 blend)
  • Longest contextHermes 3 70B InstructHermes 3 70B Instruct 131,072 · Llama 3.1 Nemotron Ultra 253B 128,000 · Aya Expanse 32B 128,000 tokens
  • Widest inputsSame inputsHermes 3 70B Instruct: Text · Llama 3.1 Nemotron Ultra 253B: Text · Aya Expanse 32B: Text
  • Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
How the score is built
MeasureWeightHermes 3 70B InstructLlama 3.1 Nemotron Ultra 253BAya Expanse 32B
Price50%5510053
Inputs & features30%0350
Context window20%323232
Specs & price only100%34/10067/10033/100

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.

Hermes 3 70B Instruct vs Llama 3.1 Nemotron Ultra 253B vs Aya Expanse 32B specifications side by side
SpecificationHermes 3 70B InstructNous ResearchLlama 3.1 Nemotron Ultra 253BNVIDIAAya Expanse 32BCohere
Capability
Capabilities Index (ECI)———
ECI rank———
Price per million tokens
Input$0.70Free (best)$0.50
Output$0.70Free (best)$1.50
Cached input———
Blended (3:1)$0.70Free (best)$0.75
Long-context rateSame rateSame rateSame rate
Price sourceMedian of 1 providersOfficial Nvidia APIOfficial Cohere API
Limits
Context window131,072 tokens (best)128,000 tokens128,000 tokens
Max output16,384 tokens (best)8,192 tokens4,000 tokens
Inputs and features
TextYesYesYes
ImagesNoNoNo
PDFsNoNoNo
AudioNoNoNo
VideoNoNoNo
ReasoningNoYesNo
Tool callingNoYesNo
Structured outputNoNoNo
Availability
WeightsOpenOpenOpenCC-BY-NC-4.0
API model ID—nvidia/llama-3.1-nemotron-ultra-253b-v1c4ai-aya-expanse-32b
API providers112 (best)
ReleasedAug 18, 2024Apr 7, 2025Oct 24, 2024
Knowledge cutoffDec 31, 2023——
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.

  • Hermes 3 70B Instruct$8.40
  • Llama 3.1 Nemotron Ultra 253BFree
  • Aya Expanse 32B$8.00
04 — Questions

Which should you choose?

Which is better: Hermes 3 70B Instruct, Llama 3.1 Nemotron Ultra 253B or Aya Expanse 32B?

There is not enough shared capability evidence to name an overall winner. The scores (Llama 3.1 Nemotron Ultra 253B 67/100, Hermes 3 70B Instruct 34/100, Aya Expanse 32B 33/100) compare the available specifications and 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, Hermes 3 70B Instruct, Llama 3.1 Nemotron Ultra 253B or Aya Expanse 32B?

Llama 3.1 Nemotron Ultra 253B is cheaper at Free input / Free output per million tokens (official Nvidia API price). Hermes 3 70B Instruct costs $0.70 input / $0.70 output per million tokens (median across 1 API provider); Aya Expanse 32B costs $0.50 input / $1.50 output per million tokens (official Cohere API price). Llama 3.1 Nemotron Ultra 253B is listed as free.

Which scores higher on benchmarks?

There is no independent benchmark that covers all three models yet. Hermes 3 70B Instruct has not been scored yet, Llama 3.1 Nemotron Ultra 253B has not been scored yet and Aya Expanse 32B has not been scored yet.

Which is better for coding?

There are no published SWE-bench Verified results for Hermes 3 70B Instruct, Llama 3.1 Nemotron Ultra 253B and Aya Expanse 32B yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that Hermes 3 70B Instruct and Aya Expanse 32B does not support tool calling, which most coding agents need.

Which has the bigger context window?

Hermes 3 70B Instruct has the largest context window at 131,072 tokens, against 128,000 for Llama 3.1 Nemotron Ultra 253B and 128,000 for Aya Expanse 32B. Maximum output per response: Hermes 3 70B Instruct up to 16,384, Llama 3.1 Nemotron Ultra 253B up to 8,192, Aya Expanse 32B up to 4,000 tokens.

Which can read images, PDFs, audio or video?

Hermes 3 70B Instruct accepts text; Llama 3.1 Nemotron Ultra 253B accepts text; Aya Expanse 32B accepts text. They handle the same number of input types.

Are any of these open source?

Yes, all three publish their weights (CC-BY-NC-4.0), so you can self-host them.

Which is newer?

Llama 3.1 Nemotron Ultra 253B is the newest, released Apr 7, 2025. Aya Expanse 32B came out Oct 24, 2024; Hermes 3 70B Instruct came out Aug 18, 2024. Knowledge cutoff: Hermes 3 70B Instruct Dec 31, 2023.

How do you decide the winner?

This rating uses price (50%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (30%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (20%, 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 Nov 18, 2026.