LFM2.5-2.6B vs Llama 3.1 70B Instruct

Prices checked

LFM2.5-2.6B takes 3 of 5 rounds against Llama 3.1 70B Instruct. It is also the cheaper of the two to run, at $0.13 per million tokens against $0.17.

Liquid AIWins 3 rounds

LFM2.5-2.6B

128k context · tool calling · open weights

$0.13/M tokens

Cheapest price, at NanoGPT · 7.2¢ per 1,000 pages

Meta

Llama 3.1 70B Instruct

131k context · tool calling · open weights

$0.17/M tokens

Cheapest price, at Hyperbolic · 9¢ per 1,000 pages

Maker's price in (lower wins)

$0.35/M

$0.40/M

Maker's price out (lower wins)

$0.50/M

$0.40/M

Cheapest price (lower wins)

$0.13/M at NanoGPT

$0.17/M at Hyperbolic

Maker's price, blended (lower wins)

$0.39/M

$0.40/M

Cost per 1,000 pages (lower wins)

7.2¢

9¢

Context

128k tokens

131k tokens

Open weights

yes

yes

Tool calling

yes

yes

Structured output

yes

yes

Reads images

no

no

Reasoning

yes

no

Released

11 Aug 2026

23 Jul 2024

Providers selling it

4

18

Price we measure against

the maker's own

OpenRouter

Trains on your prompts (no is better)

not published

not published

Providers selling both · 2

  • Kilo Gateway$0.39 / $0.40
  • OpenRouter$0.39 / $0.40

LFM2.5-2.6B, then Llama 3.1 70B Instruct, per million tokens.

Go further

Add a third or fourth model, or change the weighting, in the tool.

What moved this week

  1. Llama 3.1 70B Instruct

    Eden AI · 100% dearer ·

    $0.80 / $0.80 $0.40 / $0.40

Frequently asked

Which is cheaper, LFM2.5-2.6B or Llama 3.1 70B Instruct?
LFM2.5-2.6B, at $0.13 per million tokens against $0.17 for Llama 3.1 70B Instruct, blended 3 in : 1 out, read on 25 Oct 2026.
Where is each one cheapest?
LFM2.5-2.6B: $0.13 at NanoGPT. Llama 3.1 70B Instruct: $0.17 at Hyperbolic.
Can I buy both from one provider?
Yes. 2 providers sell both: Kilo Gateway, OpenRouter. The cheapest of them for the two together is Kilo Gateway.
Which holds more context?
Llama 3.1 70B Instruct, at 131k tokens against 128k.

Prices exclude tax, read 25 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

More