llama-3.1-70b-instruct vs sonar-pro

Capability for the money

Not the same tier: sonar-pro leads llama-3.1-70b-instruct by 16.9 points on hard reasoning (GPQA Diamond), for 6.0× the price. They are level on coding (LiveCodeBench), so the premium only pays off on hard reasoning work.

1. Coding

   writing and fixing code

   LiveCodeBench
   llama-3.1-70b-instruct23.2%

   sonar-pro27.5%

   sonar-pro +4.3

2. Hard reasoning

   graduate-level science and multi-step analysis

   GPQA Diamond
   llama-3.1-70b-instruct40.9%

   sonar-pro57.8%

   sonar-pro +16.9

Averaged over the benchmarks both models sat, llama-3.1-70b-instruct returns 32.0 points of pass rate per dollar of blended price against 7.1 — the ratio that decides whether the dearer model earns its rate.

Each figure is a pass rate: the share of that test set the model answered correctly, measured by a third party rather than self-reported by the lab. Rows are different tests of different difficulty, so compare a model against the other model, not one row against the next. No published following instructions or tool use score for one of these models, so those lenses are omitted. And a benchmark is not your workload — it narrows the shortlist, it does not pick for you.