Llama 3.2 3B Instruct vs Nex N2.5 Pro
Prices checked
Nex N2.5 Pro takes 3 of 5 rounds against Llama 3.2 3B Instruct. Llama 3.2 3B Instruct counters on price: $0.02 per million tokens against $0.12, for 131k context instead of 262k.
MetaLlama 3.2 3B Instruct
131k context · open weights
$0.02/M tokensCheapest price, at Inference · 1.3¢ per 1,000 pages
Nex N2.5 Pro
262k context · tool calling · reads images
$0.12/M tokensNex AGI's own price · 6¢ per 1,000 pages
Maker's price in (lower wins)
$0.05/M
$0.07/M
Maker's price out (lower wins)
$0.33/M
$0.25/M
Cheapest price (lower wins)
$0.02/M at Inference
none below the maker
Maker's price, blended (lower wins)
$0.12/M
$0.12/M
Cost per 1,000 pages (lower wins)
1.3¢
6¢
Context
131k tokens
262k tokens
Open weights
yes
no
Tool calling
no
yes
Structured output
yes
yes
Reads images
no
yes
Reasoning
no
yes
Released
25 Sept 2024
8 Sept 2026
Providers selling it
15
5
Price we measure against
OpenRouter
the maker's own
Trains on your prompts (no is better)
not published
no provider to check
Providers selling both · 3
NanoGPT$0.04 / $0.06
Kilo Gateway$0.12 / $0.12
OpenRouter$0.12 / $0.12
Llama 3.2 3B Instruct, then Nex N2.5 Pro, per million tokens.
Go further
Add a third or fourth model, or change the weighting, in the tool.
What moved this week
- 5¢ / $0.33 5.1¢ / $0.34
- OpenRouter · 2% cheaper · 5¢ / $0.33 5.1¢ / $0.34
Frequently asked
- Which is cheaper, Llama 3.2 3B Instruct or Nex N2.5 Pro?
- Llama 3.2 3B Instruct, at $0.02 per million tokens against $0.12 for Nex N2.5 Pro, blended 3 in : 1 out, read on 31 Oct 2026.
- Where is each one cheapest?
- Llama 3.2 3B Instruct: $0.02 at Inference. Nex N2.5 Pro: nobody sells it below the maker's own price with a stated reason, so $0.12 is the number to beat.
- Can I buy both from one provider?
- Yes. 3 providers sell both: NanoGPT, Kilo Gateway, OpenRouter. The cheapest of them for the two together is NanoGPT.
- Which holds more context?
- Nex N2.5 Pro, at 262k tokens against 131k.
Prices exclude tax, read 31 Oct 2026 from each provider's own catalog. They are per M tokens, blended 3 in : 1 out, and a page is 600 tokens in and 60 out. Bars are scaled to the larger value in each row.
Related comparisons