Calculator Models Hardware Compare

Product

Calculator
Compare
Tier List

Browse

Models
Hardware
Docs

About

Why It Works
What's New
Legal Notice
Privacy Policy

All estimates are approximations based on mathematical models and public specifications. Actual performance may vary. Do not make purchasing decisions based solely on these estimates.

Data sourced from Hugging Face, Ollama, and official model documentation. Model names and logos are trademarks of their respective owners.

© 2026 Will It Run AI — Fase Consulting Ibiza, S.L. (NIF: B57969656)

Home/Hardware/GPUs/GTX 1660 Super 6GB

NVIDIA

GTX 1660 Super 6GB

GTX 16ConsumerTuringPCIe 3CUDA

6GB

VRAM

336GB/s

Bandwidth

10TFLOPS

FP16 Compute

40TOPS

INT8 Inference

GTX 1660 Super 6GBCategory AvgIntel Arc A550M 8GB

Specifications

Compute

FP1610 TFLOPS

INT840 TOPS

ArchitectureTuring

Memory

VRAM6 GB

Bandwidth336 GB/s

General

FamilyGTX 16

SegmentConsumer

InterconnectPCIe 3

Compute PlatformCUDA

Architecture

Turing

Turing is NVIDIA's first-generation RTX architecture, introducing dedicated RT and Tensor Cores to consumer GPUs for the first time. Built on TSMC's 12nm FinFET process.

AI Relevance

The first consumer architecture with Tensor Cores, enabling meaningful acceleration for INT8 and FP16 inference. However, limited VRAM (typically 6-11 GB) restricts modern LLM model sizes.

Process: TSMC 12nmPlatform: CUDATensor Cores: Gen 2Precisions: FP32, FP16, INT8, INT4

Recommendations by Workload

Agentic Coding

B

StarCoder2 3B

This model is still usable for agentic-coding, but it is not the most specialized pick. It sits in the middle of the current model mix. It fits natively with comfortable headroom.

Decode 101.0 tok/s · 45K ctx · llama.cpp

4.3 GB / 6.0 GB VRAM

Chat

B

Qwen 3 4B

This model is a direct match for chat. It sits in the middle of the current model mix. It fits natively with comfortable headroom. Known channels: huggingface, ollama, lm-studio.

Decode 75.7 tok/s · 10K ctx · llama.cpp

4.7 GB / 6.0 GB VRAM

Coding

B

StarCoder2 3B

This model is a direct match for coding. It sits in the middle of the current model mix. It fits natively with comfortable headroom.

Decode 101.0 tok/s · 23K ctx · llama.cpp

4.1 GB / 6.0 GB VRAM

RAG

B

SmolLM3 3B

This model is still usable for rag, but it is not the most specialized pick. It sits in the middle of the current model mix. It fits natively with comfortable headroom. Known channels: huggingface, lm-studio.

Decode 101.0 tok/s · 45K ctx · llama.cpp

4.3 GB / 6.0 GB VRAM

Reasoning

B

Phi 4 Mini 4B

This model is a direct match for reasoning. It belongs to a current frontier family for local AI. It fits natively with comfortable headroom. Known channels: huggingface, ollama, lm-studio.

Decode 75.7 tok/s · 20K ctx · llama.cpp

4.7 GB / 6.0 GB VRAM

Full Model Compatibility

Llama 3.2 3B Instruct

3B4.5 GB87 tok/s22K ctx

Qwen2.5 3B Instruct

3B4.1 GB101 tok/s23K ctx

4B4.7 GB76 tok/s20K ctx

4B4.7 GB76 tok/s20K ctx

2B3.9 GB118 tok/s24K ctx

2B3.5 GB152 tok/s27K ctx

Gemmasutra Mini 2B v1

2B3.5 GB152 tok/s27K ctx

Qwen2.5 1.5B Instruct

1.5B3.2 GB185 tok/s30K ctx

Llama 3.2 1B Instruct Q8 0

1B3.1 GB194 tok/s31K ctx

TinyLlama 1.1B Chat v1.0

1.1B3.0 GB185 tok/s32K ctx

SmolVLM 500M Instruct

0.5B2.7 GB194 tok/s35K ctx

embeddinggemma 300M

0.3B2.5 GB194 tok/s38K ctx

Llama 2 7B Chat

7B6.9 GB38 tok/s14K ctx

Mistral 7B Instruct v0.2

7B6.9 GB38 tok/s14K ctx

Mistral 7B Instruct v0.3

7B6.9 GB38 tok/s14K ctx

DeepSeek R1 671B

671B416.6 GB2 tok/s4K ctx

Devstral 2 123B Instruct

123B95.7 GB3 tok/s4K ctx

744B461.6 GB2 tok/s4K ctx

27B22.2 GB11 tok/s4K ctx

Qwen3.5 35B A3B

35B28.3 GB9 tok/s4K ctx

9B8.4 GB34 tok/s11K ctx

1000B616.5 GB2 tok/s4K ctx

Mistral Large 3

675B419.7 GB2 tok/s4K ctx

Mistral Small 4 119B

119B75.1 GB7 tok/s4K ctx

Qwen3-Coder 30B A3B Instruct

30.5B20.9 GB26 tok/s5K ctx

Qwen3-Coder 480B A35B Instruct

480B299.8 GB2 tok/s4K ctx

Qwen3-Coder-Next

80B51.1 GB12 tok/s4K ctx

Qwen3.5 9B Uncensored HauhauCS Aggressive

9B8.4 GB34 tok/s11K ctx

Qwen3.5 122B A10B

122B80.3 GB3 tok/s4K ctx

Meta Llama 3.1 8B Instruct

8B7.6 GB38 tok/s13K ctx

DeepSeek V3 671B

671B416.6 GB2 tok/s4K ctx

141B93.6 GB4 tok/s4K ctx

72B56.7 GB4 tok/s4K ctx

Qwen 3 235B A22B

235B148.3 GB3 tok/s4K ctx

Qwen3-VL 30B A3B Instruct

30B20.6 GB27 tok/s5K ctx

llava llama 3 8b v1 1

8B7.6 GB38 tok/s13K ctx

Qwen3.5 397B A17B

397B305.7 GB2 tok/s4K ctx

Devstral Small 2 24B Instruct

24B19.9 GB13 tok/s5K ctx

70B55.1 GB4 tok/s4K ctx

Llama 4 Maverick 17B 128E

400B248.2 GB2 tok/s4K ctx

DeepSeek R1 0528 Qwen3 8B

8B7.6 GB38 tok/s13K ctx

111B86.6 GB3 tok/s4K ctx

Qwen 2.5 Coder 32B

32B26.0 GB10 tok/s4K ctx

Qwen 2.5 VL 72B

72B56.7 GB4 tok/s4K ctx

27B22.2 GB11 tok/s4K ctx

9B8.4 GB34 tok/s11K ctx

Meta Llama 3 8B Instruct

8B7.6 GB38 tok/s13K ctx

Qwen3.5 35B A3B

35B28.3 GB9 tok/s4K ctx

Codestral 2 25.08

22B18.4 GB14 tok/s5K ctx

Devstral Small 1.1

24B19.9 GB13 tok/s5K ctx

Just out of reach

Models you could run with an upgrade

High-quality models that need a bit more memory

DeepSeek R1 671B

671BTier 5Needs ~422.4 GB

Devstral 2 123B Instruct

123BTier 5Needs ~115.0 GB

Runs on Mac Studio M3 Ultra 256GB

744BTier 5Needs ~467.8 GB

27BTier 5Needs ~26.4 GB

Runs on RTX 5090 32GB (~$1,999)

Qwen3.5 35B A3B

35BTier 5Needs ~33.8 GB

Runs on Mac mini M4 64GB (~$1,099)

Upgrade paths

Upgrade from GTX 1660 Super 6GB

See what you unlock with more powerful hardware

Upgrade options

Upgrade options

Intel Arc A550M 8GBNext step up

Unlocks Meta Llama 3.1 8B Instruct, Llama 2 7B Chat, llava llama 3 8b v1 1+99 more

GTX 1070 Ti 8GBNVIDIA upgrade

Unlocks Meta Llama 3.1 8B Instruct, Llama 2 7B Chat, llava llama 3 8b v1 1+99 more

Intel Arc B580 12GBBest value

12 GB VRAM (+6)456 GB/s (+120)

Unlocks Qwen3.5 9B, Qwen3.5 9B Uncensored HauhauCS Aggressive, Meta Llama 3.1 8B Instruct+133 more

~$249 MSRP

AMD Instinct MI350X 288GBBiggest leap

288 GB VRAM (+282)8000 GB/s (+7664)

Unlocks Devstral 2 123B Instruct, Qwen3.5 27B, Qwen3.5 35B A3B+243 more · +1135% faster avg

~$8,000 MSRP

Compare this GPU