NVIDIA GeForce RTX 5070 Ti — Benchmarks & Specs

Bottom line: how fast is the NVIDIA GeForce RTX 5070 Ti?

SpecPicks median: 121 tokens/sec across 5 cited single-user runs of 7-9B dense models at 4-bit (Q4) under llama.cpp (range 65.5–182 tok/s), from KnightLI, llama.cpp GitHub, Hardware Corner, LocalScore (2 runs). At the $1,249.99 Amazon price of the cheapest linked listing (checked 2026-10-21 12:20 UTC; prices change), that is a SpecPicks Value Index of 96.4 tok/s per $1,000. At 1440p Ultra, the NVIDIA GeForce RTX 5070 Ti averages 123 fps in Indiana Jones and the Great Circle, per PCGamesN. For local LLM inference it generates 73.2 tokens/sec running qwen2.5:14b at q4_K_M under ollama, per ComputingForGeeks. In 3DMark Fire Strike it scores 68,741 points, per Club386. Its 16 GB of VRAM is the binding constraint for local inference: the largest model on file at 4-bit quantization on this card is meta-llama/gpt-oss-20b-MoE (20B parameters, MXFP4), generating 189.4 tokens/sec under llama.cpp, per llama.cpp GitHub.

Every figure above is a row in the tables below, and each row links out to the review or public benchmark database the number was taken from; the SpecPicks median is computed here from the runs it names, each a row in the AI table, and the per-$1,000 figure divides it by the dated listing price. SpecPicks aggregates published measurements; it does not report first-party benchmark runs.