VRAMGlass contains affiliate links. As an eBay Partner, we may be compensated if you make a purchase. As an Amazon Associate I earn from qualifying purchases.We may earn from eBay and Amazon links.Learn more

GPUs tracked64Price history73 daysPrice points (24h)30,814 pointsBiggest 24h dropCMP 170HX 64GB-10.8%Biggest 24h riseRadeon PRO W7800+33.6%Updated
$ USD

GeForce RTX 3060 12GB Price, VRAM & Local LLMs

NVIDIA's RTX 3060 12GB (2021) is a 170W entry-level card with 12GB of VRAM that handles models up to 12B, making it an easy upgrade for an older PC. Ask the seller for a screenshot, since the 8GB version has a third less bandwidth.

eBay

GeForce RTX 3060 12GB price history

Current$410-0.73%
eBay
Price updated
$771$647$523$398$274
Low $345High $700
Sep 23Oct 19Nov 14
Latest price range
$345 – $700
24-hour change
-0.73%
7-day change
-14.94%
30-day change
-10.28%
View 103 listings on eBay

As an eBay Partner, VRAMGlass may be compensated if you buy through the eBay links on this page. The price you pay stays the same.

GeForce RTX 3060 12GB VRAM & full specs

  • Memory12 GB · GDDR6
  • Bandwidth360 GB/s
  • FP16 / BF1651 TFLOPS
  • INT8101.9 TOPS
Bus width
192-bit
Architecture
Ampere (GA106)
Cores
3,584
Tensor cores
112
Ecosystem
CUDA
Interface
PCIe 4.0 x16
NVLink
Not supported
Power
170 W
Type
Consumer
Launch date
Feb 25, 2021
MSRP
$329

Best GeForce RTX 3060 12GB deals

Powered by VRAMGlass real-time price tracking — handpicking the best-value hardware deals across the web

Picked from

As an eBay Partner, VRAMGlass may be compensated if you buy through the eBay links on this page. The price you pay stays the same.

  1. UsedLenovo Nvidia GeForce RTX 3060 12GB Graphics Card GPU PC (LHR)Top Rated99.8% positive (9,192 ratings)99.8% positive (9,192)$360Buy on eBay
  2. UsedHP Nvidia RTX 3060 12GB Graphics Card GPU PC (LHR)Top Rated99.8% positive (9,192 ratings)99.8% positive (9,192)$360Buy on eBay
  3. UsedHP nVidia RTX 3060 12GB GDDR6 GPU Graphics Card TPC-M011G100% positive (1,135 ratings)100% positive (1,135)$369Buy on eBay
  4. UsedPNY GeForce RTX 3060 XLR8 Gaming REVEL EPIC-X RGB Single Fan 12GB GDDR6 Graphics100% positive (1,068 ratings)100% positive (1,068)$370Buy on eBay
  5. UsedAsus GeForce RTX 3060 ROG Strix Gaming 12GB GDDR6 Graphics CardTop Rated99.7% positive (2,150 ratings)99.7% positive (2,150)$375Buy on eBay
  6. UsedHP NVIDIA GeForce RTX 3060 12GB GDDR6 Graphics Card GPU - OEM Pull99.7% positive (4,692 ratings)99.7% positive (4,692)$379Buy on eBay
See more on eBay
Picked from

As an Amazon Associate I earn from qualifying purchases. The price you pay stays the same.

  1. UsedDual NVIDIA GeForce RTX 3060 V2 OC Edition 12GB GDDR6 Gaming Graphics Card (PCIe 4.0, 12GB GDDR6 Memory, HDMI 2.1, DisplayPort 1.4a, 2-Slot Design, Axial-tech Fan Design, 0dB Tech) (Renewed)5 stars (2 ratings)5 stars (2)$410$450-9%Buy on Amazon
  2. UsedGaming GeForce RTX 3060 12GB GDDR6 PCI Express 4.0 ATX Video Card GV-N3060GAMING OC-12GD (Renewed)4.4 stars (127 ratings)4.4 stars (127)$420Buy on Amazon
  3. UsedGeForce RTX 3060 XC Gaming, 12G-P5-3657-KR, 12GB GDDR6, Dual-Fan, Metal Backplate (Renewed)4.2 stars (35 ratings)4.2 stars (35)$420$500-16%Buy on Amazon
See more on Amazon

What LLMs can a single GeForce RTX 3060 12GB run?

One RTX 3060 12GB is best with up to 12B at 4-bit, Gemma 4 12B at about 36 tokens/s. 30B-class MoE runs at about 45 with enough system RAM.

ModelSizeRuns?Speed t/sMax context
Gemma 4 E4B8B5 GBRuns50.5128K
Qwen3.5 9B9.7B5.7 GBRuns44.4182K
Gemma 4 12B12B7.1 GBRuns35.8256K
Gemma 4 26B A4B25.2B (3.8B active)16.9 GBNeeds RAM offload33.6256K
Qwen3.6 27B27.8B16.8 GBToo big——
Qwen3.8 27B27.8B16.5 GBToo big——
GLM 4.7 Flash30B (3B active)18.3 GBNeeds RAM offload33.9180K
Gemma 4 31B30.7B18.3 GBToo big——
Qwen3.6 35B A3B36B (3B active)22.1 GBNeeds RAM offload45.4256K

RunsFits fully in VRAM; speed is estimated from VRAM bandwidth.

Needs RAM offloadMoE only: expert weights sit in system RAM, so speed is estimated from 70 GB/s RAM bandwidth.

Too bigWon’t fit on one card at this quantization, even with RAM offload.

Full calculation rules are on the Methodology page.

What can the GeForce RTX 3060 12GB run with Strata?

With Strata, the GeForce RTX 3060 12GB can run models far larger than its VRAM. Strata keeps most of the model in RAM and on the SSD and only the most-used experts on the GPU. Benchmarks come from Strata’s author and community, with multi-GPU setups marked by card count and today’s prices added by VRAMGlass.

Where Strata puts the model

Qwen3.8 Flash Next Q2_0 · 66.4 GB
  • VRAM 12 GB: the parts every token uses, plus the most-used experts
  • RAM about 40 GB: all the experts; the CPU computes the ones the GPU doesn’t hold
  • SSD 28.8 GB: the lookup table, a few rows read per token

Qwen3.8 Flash Next

IQ3_XXS · 75.8 GBMeasured
Output55.4tokens/s
Prefill962tokens/s
  • GPUGeForce RTX 3060 12GB$410
  • RAM46.2 GiB
  • CPUIntel Core Ultra 7 265
  • StorageLexar NM1090 PRO 2 TB

Qwen3.8 Flash Next

IQ2_XS · 68 GBMeasured
Output25.8–26.5tokens/s
Prefill177tokens/s
  • GPUGeForce RTX 3060 12GB$410
  • RAM251.8 GiB DDR3
  • CPU2 × Intel Xeon E5-2630 v2
  • StorageSolidigm NVMe

Qwen3.8 Flash Next

IQ2_XS · 68 GBMeasured
Output45.5tokens/s
Prefill829tokens/s
  • GPUGeForce RTX 3060 12GB$410
  • RAM2 × 32 GB DDR4-3200$484
  • CPUIntel Core i7-12700
  • Storage2 TB NVMe

Qwen3.8 Flash Next

Q2_0 · 66.4 GBMeasured
Output42.9tokens/s
Prefill882tokens/s
  • GPUGeForce RTX 3060 12GB$410
  • RAM2 × 32 GB DDR4-3200$484
  • CPUIntel Core i7-12700
  • Storage2 TB NVMe

Qwen3.8 Flash Next

IQ3_XXS · 75.8 GBMeasured
Output26.2tokens/s
Prefill681tokens/s
  • GPUGeForce RTX 3060 12GB$410
  • RAM2 × 32 GB DDR4-3200$484
  • CPUIntel Core i7-12700
  • Storage2 TB NVMe

Qwen3.8 Flash Next

IQ3_S · 83.6 GBMeasured
Output20.2tokens/s
Prefill569tokens/s
  • GPUGeForce RTX 3060 12GB$410
  • RAM2 × 32 GB DDR4-3200$484
  • CPUIntel Core i7-12700
  • Storage2 TB NVMe

Qwen3.8 Flash Next

IQ3_S · 83.6 GBMeasured
Output43.6tokens/s
Prefill603tokens/s
  • GPU2 × GeForce RTX 3060 12GB$820
  • RAM121.5 GiB DDR3-1867
  • CPU2 × Intel Xeon E5-2678 v3
  • Storage1.4 TB NVMe

As an eBay Partner, VRAMGlass may be compensated if you buy through the eBay links on this page. The price you pay stays the same.

Who it suits

Good fit

  1. Running models up to 12B

    Gemma 4 12B and Qwen3.5 9B fit entirely at 4-bit, at about 36 to 44 tokens/s, with over 180K of context

  2. You want to try 30B-class MoE models

    Qwen3.6 35B A3B at 4-bit with experts in system RAM runs at about 45 tokens/s, faster than Gemma 4 12B running fully on the card

  3. Upgrading an older PC

    The card draws 170W, NVIDIA recommends a 550W power supply, and it gets a full 16 lanes even on an older PCIe 3.0 motherboard

  4. You want to run Qwen3.8 Flash Next with Strata

    12GB is exactly its minimum VRAM, and with 64GB of system RAM every version runs (Strata hasn't published 3060 results yet)

Not a good fit

  1. Generation speed matters to you

    On the same Gemma 4 12B, the RTX 5060 Ti 16GB does about 44 and the RTX 5070 about 64 tokens/s

  2. You want a 12B at 8-bit

    Gemma 4 12B at 8-bit needs 12.7 GB, just short of fitting

  3. You want 27B to 31B dense models

    They all need 16.5 GB or more at 4-bit; only Qwen3.8 27B at 2-bit fits, with 29K of context at about 26 tokens/s

Before you buy

  1. Make sure it's the 12GB desktop version

    There's also an 8GB RTX 3060, with the memory bus cut from 192-bit to 128-bit and a third less bandwidth; the higher-tier RTX 3060 Ti also has only 8GB. If a listing just says “3060”, ask the seller for a GPU-Z screenshot to confirm 12GB of VRAM and a 192-bit bus.

  2. No Founders Edition

    NVIDIA didn't make a Founders Edition 3060, so every card on the market comes from a partner brand; check length, thickness and power connectors for the specific model.

  3. Full lanes even on older motherboards

    The 3060 is wired for all 16 PCIe lanes, so on an older PCIe 3.0 motherboard it still gets 3.0 x16. The RTX 4060 Ti 16GB and RTX 5060 Ti 16GB use only 8 lanes, so on the same board they get half the 3060's bandwidth. Once a model is fully in VRAM, PCIe mainly affects loading speed.

  4. Check your system RAM before running MoE models

    A 12GB card runs 30B-class MoE models by holding experts in system RAM. Qwen3.6 35B A3B at 4-bit totals 22.1 GB, and whatever doesn't fit in VRAM sits in RAM, so if you don't have enough RAM, it won't start. Generation speed then also depends on system memory bandwidth, so DDR5 platforms are faster than DDR4.

Compared with similar GPUs

GeForce RTX 3060 12GB vs GeForce RTX 5070, GeForce RTX 4060 Ti 16GB, and GeForce RTX 2080 Ti 22GB: price, VRAM, bandwidth, compute and local LLM speed compared.

SpecGeForce RTX 3060 12GBGeForce RTX 5070GeForce RTX 4060 Ti 16GBGeForce RTX 2080 Ti 22GB
Memory
12 GB
12 GBSame
16 GB33% more
22 GB83% more
Bandwidth
360 GB/s
672 GB/s87% higher
288 GB/s20% lower
616 GB/s71% higher
BF16 / FP16 compute
51 TFLOPS
124 TFLOPS142% higher
88 TFLOPS73% higher
108 TFLOPS111% higher
INT8 compute
102 TOPS
247 TOPS142% higher
177 TOPS73% higher
215 TOPS111% higher
Qwen3.8 27B speed (2-bit)
26.4 t/s
47.8 t/s81% faster
21.3 t/s19% slower
44 t/s67% faster
Largest dense model (4-bit)
16B
16BSame
23B44% more
34B113% more
Power
170 W
250 W47% higher
165 WSame
250 W47% higher
GPU interconnect
Not supported
Not supported
Not supported
NVLink 100 GB/s
Interface
PCIe 4.0 x16
PCIe 5.0 x16
PCIe 4.0 x8
PCIe 3.0 x16
Released
2021-02
2025-03
2023-07
2018-09
eBay median
$410Buy GeForce RTX 3060 12GB
$925Buy GeForce RTX 5070
$749Buy GeForce RTX 4060 Ti 16GB
$551Buy GeForce RTX 2080 Ti 22GB

As an eBay Partner, VRAMGlass may be compensated if you buy through the eBay links on this page. The price you pay stays the same.

Other markets

Amazon.jp · Japan
$499
26 listings
View
Xianyu · Mainland China
$319
16 listings
View