ZMIME
Comparison · 3 models · Updated Nov 18, 2026

Qwen2.5 14B Instruct vs Nova Lite 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

Retired Qwen2.5 14B Instruct is no longer offered by any provider we track. Its figures are the last published listing, kept so this comparison stays available.

  1. Alibaba (Qwen)

    Qwen2.5 14B Instruct

    Released Sep 2024Retired

    42/100
    • ECI—
    • Price$0.35 / $1.40
    • Context131K
  2. Amazon

    Nova Lite

    Released Dec 3, 2024

    76/100
    • ECI—
    • Price$0.06 / $0.24
    • Context300K
  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 (Nova Lite 76/100, Qwen2.5 14B Instruct 42/100, 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 priceNova LiteNova Lite $0.105 · Qwen2.5 14B Instruct $0.613 · Aya Expanse 32B $0.75 per 1M tokens (3:1 blend)
  • Longest contextNova LiteNova Lite 300,000 · Qwen2.5 14B Instruct 131,072 · Aya Expanse 32B 128,000 tokens
  • Widest inputsNova LiteQwen2.5 14B Instruct: Text · Nova Lite: Text, Images, PDFs, Video · Aya Expanse 32B: Text
  • Self-hostingQwen2.5 14B Instruct and Aya Expanse 32BPublishes downloadable weights (CC-BY-NC-4.0)
How the score is built
MeasureWeightQwen2.5 14B InstructNova LiteAya Expanse 32B
Price50%609253
Inputs & features30%25700
Context window20%244532
Specs & price only100%42/10076/10033/100

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.

Qwen2.5 14B Instruct vs Nova Lite vs Aya Expanse 32B specifications side by side
SpecificationQwen2.5 14B InstructAlibaba (Qwen)Nova LiteAmazonAya Expanse 32BCohere
Capability
Capabilities Index (ECI)———
ECI rank———
Price per million tokens
Input$0.35$0.06 (best)$0.50
Output$1.40$0.24 (best)$1.50
Cached input—$0.015—
Blended (3:1)$0.613$0.105 (best)$0.75
Long-context rateSame rateSame rateSame rate
Price sourceOfficial Alibaba APIOfficial Amazon Bedrock APIOfficial Cohere API
Limits
Context window131,072 tokens300,000 tokens (best)128,000 tokens
Max output8,192 tokens10,000 tokens (best)4,000 tokens
Inputs and features
TextYesYesYes
ImagesNoYesNo
PDFsNoYesNo
AudioNoNoNo
VideoNoYesNo
ReasoningNoNoNo
Tool callingYesYesNo
Structured outputNoNoNo
Availability
WeightsOpenProprietaryOpenCC-BY-NC-4.0
API model IDqwen2-5-14b-instructamazon.nova-lite-v1:0c4ai-aya-expanse-32b
API providers18 (best)2
ReleasedSep 2024Dec 3, 2024Oct 24, 2024
Knowledge cutoffApr 2024Oct 2024—
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.

  • Qwen2.5 14B Instruct$6.30
  • Nova Lite$1.08
  • Aya Expanse 32B$8.00
04 — Questions

Which should you choose?

Which is better: Qwen2.5 14B Instruct, Nova Lite or Aya Expanse 32B?

There is not enough shared capability evidence to name an overall winner. The scores (Nova Lite 76/100, Qwen2.5 14B Instruct 42/100, 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, Qwen2.5 14B Instruct, Nova Lite or Aya Expanse 32B?

Nova Lite is cheaper at $0.06 input / $0.24 output per million tokens (official Amazon Bedrock API price). Qwen2.5 14B Instruct costs $0.35 input / $1.40 output per million tokens (official Alibaba API price); Aya Expanse 32B costs $0.50 input / $1.50 output per million tokens (official Cohere API price). At a typical mix of three input tokens to one output token, that is $0.105 per million tokens for Nova Lite versus $0.613 for Qwen2.5 14B Instruct (5.8× as much) and $0.75 for Aya Expanse 32B (7.1× as much).

Which scores higher on benchmarks?

There is no independent benchmark that covers all three models yet. Qwen2.5 14B Instruct has not been scored yet, Nova Lite 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 Qwen2.5 14B Instruct, Nova Lite 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 Aya Expanse 32B does not support tool calling, which most coding agents need.

Which has the bigger context window?

Nova Lite has the largest context window at 300,000 tokens, against 131,072 for Qwen2.5 14B Instruct and 128,000 for Aya Expanse 32B. Maximum output per response: Qwen2.5 14B Instruct up to 8,192, Nova Lite up to 10,000, Aya Expanse 32B up to 4,000 tokens.

Which can read images, PDFs, audio or video?

Qwen2.5 14B Instruct accepts text; Nova Lite accepts text, images, PDFs and video; Aya Expanse 32B accepts text. Nova Lite handles the widest range of inputs.

Are any of these open source?

Qwen2.5 14B Instruct and Aya Expanse 32B publishes its weights (CC-BY-NC-4.0) and can be self-hosted; Nova Lite is proprietary.

Which is newer?

Nova Lite is the newest, released Dec 3, 2024. Aya Expanse 32B came out Oct 24, 2024; Qwen2.5 14B Instruct came out Sep 2024. Knowledge cutoff: Qwen2.5 14B Instruct Apr 2024, Nova Lite Oct 2024.

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.