ZMIME
Comparison · 3 models · Updated Nov 18, 2026

Nemotron 3 Nano 30B A3B vs LightOnOCR 2 1B vs Ministral 3 3B

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. NVIDIA

    Nemotron 3 Nano 30B A3B

    Released Dec 15, 2025

    65/100
    • ECI—
    • Price$0.06 / $0.24
    • Context262K
  2. LightOn

    LightOnOCR 2 1B

    Released Jan 16, 2026

    52/100
    • ECI—
    • Price$0.101 / $0.209
    • Context16K
  3. Mistral AI

    Ministral 3 3B

    Released Dec 2, 2025

    73/100
    • ECI—
    • Price$0.10 / $0.10
    • Context262K
01 — Verdict

Capability evidence is incomplete

There is not enough shared capability evidence to name an overall winner. The scores (Ministral 3 3B 73/100, Nemotron 3 Nano 30B A3B 65/100, LightOnOCR 2 1B 52/100) 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 priceMinistral 3 3BMinistral 3 3B $0.10 · Nemotron 3 Nano 30B A3B $0.105 · LightOnOCR 2 1B $0.128 per 1M tokens (3:1 blend)
  • Longest contextNemotron 3 Nano 30B A3B and Ministral 3 3BNemotron 3 Nano 30B A3B 262,144 · Ministral 3 3B 262,144 · LightOnOCR 2 1B 16,384 tokens
  • Widest inputsLightOnOCR 2 1B and Ministral 3 3BNemotron 3 Nano 30B A3B: Text · LightOnOCR 2 1B: Text, Images · Ministral 3 3B: Text, Images
  • Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
How the score is built
MeasureWeightNemotron 3 Nano 30B A3BLightOnOCR 2 1BMinistral 3 3B
Price50%928893
Inputs & features30%352560
Context window20%43043
Specs & price only100%65/10052/10073/100

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.

Nemotron 3 Nano 30B A3B vs LightOnOCR 2 1B vs Ministral 3 3B specifications side by side
SpecificationNemotron 3 Nano 30B A3BNVIDIALightOnOCR 2 1BLightOnMinistral 3 3BMistral AI
Capability
Capabilities Index (ECI)———
ECI rank———
Price per million tokens
Input$0.06 (best)$0.101$0.10
Output$0.24$0.209$0.10 (best)
Cached input——$0.01
Blended (3:1)$0.105$0.128$0.10 (best)
Long-context rateSame rateSame rateSame rate
Price sourceMedian of 14 providersMedian of 2 providersOfficial Mistral API
Limits
Context window262,144 tokens (best)16,384 tokens262,144 tokens (best)
Max output262,144 tokens—262,144 tokens
Inputs and features
TextYesYesYes
ImagesNoYesYes
PDFsNoNoNo
AudioNoNoNo
VideoNoNoNo
ReasoningYesNoNo
Tool callingYesNoYes
Structured outputNoNoYes
Availability
WeightsOpenOpenapache-2.0OpenApache 2.0
API model ID——ministral-3b-2512
API providers14 (best)211
ReleasedDec 15, 2025Jan 16, 2026Dec 2, 2025
Knowledge cutoffJun 25, 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.

  • Nemotron 3 Nano 30B A3B$1.08
  • LightOnOCR 2 1B$1.43
  • Ministral 3 3B$1.20
04 — Questions

Which should you choose?

Which is better: Nemotron 3 Nano 30B A3B, LightOnOCR 2 1B or Ministral 3 3B?

There is not enough shared capability evidence to name an overall winner. The scores (Ministral 3 3B 73/100, Nemotron 3 Nano 30B A3B 65/100, LightOnOCR 2 1B 52/100) 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, Nemotron 3 Nano 30B A3B, LightOnOCR 2 1B or Ministral 3 3B?

Ministral 3 3B is cheaper at $0.10 input / $0.10 output per million tokens (official Mistral API price). Nemotron 3 Nano 30B A3B costs $0.06 input / $0.24 output per million tokens (median across 14 API providers); LightOnOCR 2 1B costs $0.101 input / $0.209 output per million tokens (median across 2 API providers). At a typical mix of three input tokens to one output token, that is $0.10 per million tokens for Ministral 3 3B versus $0.105 for Nemotron 3 Nano 30B A3B (1× as much) and $0.128 for LightOnOCR 2 1B (1.3× as much).

Which scores higher on benchmarks?

There is no independent benchmark that covers all three models yet. Nemotron 3 Nano 30B A3B has not been scored yet, LightOnOCR 2 1B has not been scored yet and Ministral 3 3B has not been scored yet.

Which is better for coding?

There are no published SWE-bench Verified results for Nemotron 3 Nano 30B A3B, LightOnOCR 2 1B and Ministral 3 3B yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. Note that LightOnOCR 2 1B does not support tool calling, which most coding agents need.

Which has the bigger context window?

Nemotron 3 Nano 30B A3B and Ministral 3 3B have the largest context windows (262,144 and 262,144 tokens), against 16,384 for LightOnOCR 2 1B. Maximum output per response: Nemotron 3 Nano 30B A3B up to 262,144, Ministral 3 3B up to 262,144 tokens.

Which can read images, PDFs, audio or video?

Nemotron 3 Nano 30B A3B accepts text; LightOnOCR 2 1B accepts text and images; Ministral 3 3B accepts text and images. LightOnOCR 2 1B handles the widest range of inputs.

Are any of these open source?

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

Which is newer?

LightOnOCR 2 1B is the newest, released Jan 16, 2026. Nemotron 3 Nano 30B A3B came out Dec 15, 2025; Ministral 3 3B came out Dec 2, 2025. Knowledge cutoff: Nemotron 3 Nano 30B A3B Jun 25, 2025.

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.