← All GPUs
NVIDIA · Consumer · Ada Lovelace (AD106)
NVIDIA GeForce RTX 4060 Ti 16GB
128-bit GDDR6; TGP 165 W for the 16 GB model per NVIDIA's '16 GB or 8 GB' / '165 or 160' spec rows.
VRAM
16 GB
GDDR6
Memory bandwidth
not published
FP16 compute
not published
Price
$499
launch price; sold used now, used price not tracked yet
What fits at 4-bit
The 4-bit file plus 8,192 tokens of context and 1.5 GB of overhead, against the VRAM of one card. Speeds are measured medians where someone has published a run at that card count; a blank speed means no run, not a slow one. Machine pages carry estimates for the gaps.
| Model | Needs | 1 card (16 GB) |
|---|---|---|
| Qwen3.5 4B4B · Q4_K_M | 4.3 GB | fits |
| Llama 3.1 8B Instruct8B · Q4_K_M | 7.1 GB | 48 tok/smedian of 1 run |
| Qwen3 8B8.2B · Q4_K_M | 7.3 GB | 46 tok/smedian of 1 run |
| Gemma 4 E4B IT8B · Q4_K_M | 6.5 GB | fits |
| Granite 4.2 8B8.8B · Q4_K_M | 7.9 GB | fits |
| Qwen3.5 9B9B · Q4_K_M | 7.0 GB | fits |
| Gemma 4 12B IT11.95B · Q4_K_M | 8.2 GB | fits |
| Ministral 3 14B Reasoning (2512)13.9B · Q4_K_M | 10.4 GB | fits |
| Phi-414.7B · Q4_K_M | 11.3 GB | fits |
| gpt-oss-20b21B, 3.6B active · MXFP4 (native) | 12.5 GB | 77 tok/smedian of 2 runs |
Measured speeds
One card
| Model | Quant | Backend | Generation | Prompt | Context | Source |
|---|---|---|---|---|---|---|
| gpt-oss-20b | MXFP4 MoE | llama.cpp b6471 CUDA | 91 tok/s | 3,982 tok/s2,048-token prompt | 0 | llama.cpp benchmark discussion2025-09 |
| gpt-oss-20b | MXFP4 | llama.cpp (build not stated) | 63 tok/s | 3,274 tok/s4,096-token prompt | 4,096 | review2026-03 |
| Llama 3.1 8B Instruct | Q4_K_M | LocalScore (llamafile; version not stated) | 48 tok/s | 2,233 tok/s | 0 | LocalScore2026-10 |
| Qwen3 8B | Q4_K | llama.cpp (build not stated) | 46 tok/s | 2,675 tok/s4,096-token prompt | 4,096 | review2026-03 |
Speculative decoding, several users, or long context
Kept apart because each measures a different workload from one person chatting near the start of a conversation.
| Model | Quant | Backend | Generation | Prompt | Context | Measured on | Source |
|---|---|---|---|---|---|---|---|
| gpt-oss-20b | MXFP4 | llama.cpp (build not stated) | 58 tok/s | 2,553 tok/s16,384-token prompt | 16,384 | RTX 4060 Ti 16GB | review2026-03 |
| gpt-oss-20b | MXFP4 | llama.cpp (build not stated) | 52 tok/s | 1,965 tok/s32,768-token prompt | 32,768 | RTX 4060 Ti 16GB | review2026-03 |
| Qwen3 8B | Q4_K | llama.cpp (build not stated) | 34 tok/s | 1,481 tok/s16,384-token prompt | 16,384 | RTX 4060 Ti 16GB | review2026-03 |
| Qwen3 8B | Q4_K | llama.cpp (build not stated) | 26 tok/s | 760 tok/s32,768-token prompt | 32,768 | RTX 4060 Ti 16GB | review2026-03 |
Specifications
- Memory
- 16 GB GDDR6
- FP16 basis
- NVIDIA publishes only 353 AI TOPS (sparse) for the 4060 Ti; no dense FP16 tensor figure. Bandwidth not stated on NVIDIA's page.
- Board power
- 165 W
- PCIe
- Gen 4
- NVLink
- No
- Cooling
- Open-air
- Software
- CUDA, Vulkan
- Released
- 2023
- Availability
- used only
- Price basis
- Superseded by the RTX 5060 Ti; not sold by NVIDIA.
- Maker page
- nvidia.com
Sources
- nvidia.com: 16 GB or 8 GB GDDR6, 128-bit, TGP 165 or 160 W, PCIe Gen 4
- nvidia.com: RTX 4060 Ti 16GB from $499 (July 2023)