Skip to content
ModelCap

Model decision surface

Compare AI models

Start with Nova 2 Lite and Gemma 3 12B, or choose any two current ranked language models. Compare capability evidence, price, context, provider availability, and weight access without pretending one field decides every use case.

Current public data

Nova 2 Lite vs Gemma 3 12B

Live dataset updated 9/9/2026, 8:46:14 PM UTC

Open 1200×630 evidence receipt
Factual comparison of Nova 2 Lite and Gemma 3 12B
Field
ModelCap position#129#133
Index score30.930.0
EvidenceMeasured2 public benchmark observations across 2 boardsMeasured3 public benchmark observations across 3 boards
Input / 1M$0.30$0.05
Output / 1M$2.50$0.15
Pricing statusfreshfresh
Context1M131K
Providers11
Weight accessAPI onlyGated access

Decision facts

  • Nova 2 Lite is #129; Gemma 3 12B is #133 on the same current language board.
  • Index scores are 30.9 for Nova 2 Lite and 30.0 for Gemma 3 12B. Their published uncertainty intervals overlap, so the rank difference alone does not establish a reliable capability advantage for your workload.
  • Both positions use Measured evidence.
  • Listed output price per 1M tokens is $2.50 for Nova 2 Lite and $0.15 for Gemma 3 12B. For 1,000 requests with 2,000 input and 500 output tokens each (2M input + 0.5M output), the listed-rate estimate is $1.85 for Nova 2 Lite versus $0.17 for Gemma 3 12B. Gemma 3 12B costs 90.5% less in this scenario. This excludes caching, batch discounts, prompt-length tiers, tool charges and retries; verify the selected endpoint before budgeting.
  • Published context is 1,000,000 tokens for Nova 2 Lite and 131,072 for Gemma 3 12B.
  • Weight access differs: Nova 2 Lite is none; Gemma 3 12B is gated.

These are separate published fields, not a synthetic winner. ModelCap does not collapse price, access, context, and capability evidence into a hidden recommendation score.

Each comparison page is a permanent, shareable URL with the same live figures as this tool: ModelCap Index position, API pricing, context window, provider count, weight access and every shared benchmark board.