NVIDIA GeForce RTX 4070 Ti Super

Ada's sweet spot: 672 GB/s and 16 GB — 13B at Q6, IQ8 and MoE 30B on the fastest 16 GB bus of the generation, at 285 W.

16 GB GDDR6X 672 GB/s 256-bit bus 285 W TDP

Technical specifications

GPU die
AD103
CUDA cores
8448
Boost clock
~2.61 GHz
VRAM
16 GB GDDR6X
Memory bus
256-bit
Memory bandwidth
672 GB/s
PCIe
4.0 x16
NVLink
none
TDP
285 W
Release
Jan 2024

What these specs mean for llama.cpp

The best of both Ada halves

The Ti Super is the only Ada card that pairs a 16 GB pool with a fast GDDR6X bus. What that combination buys, at 672 GB/s:

Token speed tracks the 4070 Ti's bus; the VRAM is the upgrade. Prompt processing (~900–1200 t/s on 13B) is the best in the 16 GB tier.

What it still can't do

27B Q4 needs 19 GB — 3 GB over. 13B plain Q8_0 (17.4 GB) misses by 1.4 GB. Those are the two lines this card sits just below, and a pair crosses both.

Why it's the Ada pick The 4060 Ti 16G has the VRAM at half the speed; the 4080 adds ~7% more bandwidth for roughly twice the money. The Ti Super is where VRAM and speed agree at one price point: it is this generation's single-card recommendation for 13B quality and MoE work.

The MTP question: if your model ships a built-in multi-token-prediction head (Qwen 3.5/3.6/3.8, Gemma 4), the one-flag speculative pass (--spec-type draft-mtp --spec-draft-n-max 2 --parallel 1) is worth +45–65% (estimated) on token generation here; no per-card A/B in the record; the estimate matches its 672 GB/s bus class at n-max 2. The price is ~0.6–2 GB of VRAM, the output is bit-identical to the unaccelerated run, and models without MTP heads (Llama class) see nothing. Full tuning rules in the MTP guide.

Expected performance (Q4_K_M, single GPU)

ModelSize on GPUToken genToken gen (MTP est.)Prompt proc.Fits?
8B Q8_0~9.7 GB~75–100 t/s~109–165 t/s~900–1200 t/s✅ 32K+ ctx
13B Q4_K_M~8.5 GB~45–58 t/s~65–95 t/s~900–1200 t/s✅ 32K+ ctx
13B Q6_K~13.8 GB~32–42 t/s~46–69 t/s~750–950 t/s✅ 8–16K ctx
27B Q3_K_M~15.8 GB~11–15 t/s~16–25 t/s~500–650 t/s⚠️ 4–8K ctx
MoE 30B-A3B (Q3)~14.5 GB~32–42 t/s~46–70 t/s~800–1000 t/s⚠️ 4–8K ctx, 3B active

MTP est. = rough prior with --spec-type draft-mtp for models that ship MTP heads (Qwen 3.5/3.6/3.8, Gemma 4); Llama-class models get no MTP speedup. Actual gains depend on model, quant, context length, llama.cpp build and your card — sweep --spec-draft-n-max, don't trust the column blindly. MTP guide →

Figures assume full offload (-ngl 99), ~8K context and batch 2048. Token speed tracks the memory bus almost linearly, so treat ranges as class estimates — your llama.cpp build and the model's attention layout shift them. The Q6_K and IQ8_0 lines are what the extra 4 GB over the 4070 Ti actually buys.

27B models (Qwen class) — Q4 / Q5 / Q6 / Q8

QuantGGUF size1× 4070 Ti S (16 GB)2× 4070 Ti S (32 GB)
Q3_K_M~15.8 GB⚠️ ~11–15 t/s, 4–8K ctx✅ (overkill)
Q4_K_M~19 GB❌ ~3 GB over✅ ~22–28 t/s, 8–16K ctx
Q5_K_M~22 GB❌✅ ~19–24 t/s, 8K ctx
Q6_K~25 GB❌✅ ~16–21 t/s, 8K ctx
Q8_0~33 GB❌❌ ~1 GB over

Single-card 27B is Q3 at short context — the honest ceiling. The pair (570 W, 32 GB) is Ada's best single-rig 27B story: Q4 at 8–16K, Q5/Q6 at 8K, and 13B at full Q8_0 (17.4 GB) with room to spare. It sits ~40% behind a 2× 3090 NVLink rig at the same models — and ~3× behind a 2× 5070 Ti — but it's the fastest 16 GB pair of its generation at street prices.

Practical tips

Multi-card: 2×, 3×, 4×

PCIe 4.0, ~25–35% tax. 16 GB tiles: the pair lands on the 27B Q4 line with 32 GB.

BuildGPU powerVRAM13B Q8_027B Q4_K_M27B Q5_K_M
2× 4070 Ti S570 W32 GB~55–70 t/s, 16K+ ctx✅ ~22–28 t/s, 8–16K ctx~19–24 t/s, 8K ctx

Two slots on a normal ATX board close the job. At 570 W total it's a serious rig — but it fits in a regular case with a 2×8-pin-800W PSU.

Where it sits vs 2× 3090 (NVLink, 48 GB): the 3090 pair wins on speed per token (no split tax) and VRAM — the default 70B answer. vs 2× 5070 Ti (PCIe 5.0, 32 GB): the Blackwell pair is ~3× faster per token. The Ti Super pair is the value-speed point: the fastest Ada pair you can buy at street money.

Offloading (CPU RAM)

When a model won't fit, -ngl N parks the tail layers on the CPU. The mental model that saves pain: offloaded layers don't run at a fraction of GPU speed — they run at CPU speed. Handy moves: peel the last 2–4 layers off to reclaim ~0.5–1 GB of VRAM for almost nothing; keep MoE experts on CPU with --n-cpu-moe; park the output embedding with -ot output=CPU on big models. Budget 32 GB system RAM for 27B-class offload, 64 GB for 70B. And never offload the KV cache — the session becomes unusable.

2-card example llama-server -m 27b-q4_k_m.gguf -ngl 99 -sm layer -c 16384 --tensor-split 1,1

Recommended llama.cpp command

llama-server -m llama-3.1-13b-q6_k.gguf -ngl 99 -c 16384 -b 2048 -ub 2048 --host 127.0.0.1 --port 8080

Verdict

The 4070 Ti Super is the card the 40 series should have started with: 16 GB and 672 GB/s in one envelope, at 285 W. It's the generation's single-card answer for 13B quality work and MoE models, and — in pairs — its 27B answer. It loses to the 3090 on speed-per-token and to the 50-series on everything, but at street prices it's the best Ada LLM card made.

← RTX 4070 Ti next: RTX 4080 →