Consumer GPU
NVIDIA GeForce RTX 5090 32GB
8B–32B quantized models with room reserved for runtime and KV cache.
MyAI rating
Not scored
N/A
VRAM
32 GB
TDP
575 W
Best LLM
No benchmark
MSRP
$2.0k
Curated Aggregate·2026-08-1030 workloads / Blackwell
Benchmarked GPU index
The GPU pagebuyers can actually use.
Real GPUs. Real benchmark coverage. Real next clicks. This page now surfaces your actual AI hardware catalog and sends every serious card into its own benchmark detail page.
GPUs surfaced
43
Consumer, pro, and datacenter cards
Benchmark records
371
Linked back to methodology-aware detail pages
Average rating
Not scored
Composite scoring from throughput, trust, and coverage
Top consumer pick
5090 32GB
Editorial + benchmark-backed
Fresh from your rig
AMD Radeon RX 7900 XTX 24GB local Ollama sweep
202.10 tok/s on llama3.2:3b / 115.60 tok/s on qwen2.5:7b. Captured on 2026-06-01 with ollama-local-api.
Pro GPU
AMD Radeon Pro W7900 48GB
Benchmark-backed GPU profile with hardware detail, trust signals, and buying path.
MyAI rating
Not scored
N/A
VRAM
48 GB
TDP
295 W
Best LLM
No benchmark
MSRP
$4.0k
Curated Aggregate·2025-08-062 workloads / RDNA 3
Datacenter GPU
NVIDIA H100 SXM5 80GB
70B quantized inference. 70B FP16 and 405B Q4 require more memory than one 80GB device.
MyAI rating
Not scored
N/A
VRAM
80 GB
TDP
700 W
Best LLM
No benchmark
MSRP
$25k
Curated Aggregate·2026-04-0913 workloads / Hopper
Quick picks
Three routes through the market
Best local 70B path
4× NVIDIA RTX 3090 24GB
Prioritize VRAM headroom and clean 32B-to-70B scaling.
Not scoredNo benchmark
Best value class
NVIDIA GeForce RTX 3090 24GB
The used-market sweet spot still matters for serious builders.
Not scoredNo benchmark
Best scale-up card
NVIDIA H100 SXM5 80GB
When this turns into serving, capacity and interconnect win.
Not scoredNo benchmark
Top LLM throughput
Best benchmarked token movers
Browse hardwares
Real GPUs, fully clickable
Search by name, slice by class, then jump straight into the hardware brief. These cards are built from your benchmark records, not a static mock list.
Showing 43 GPUs / All classes / all brands.
Datacenter GPUNVIDIA
NVIDIA H100 SXM5 80GB
Rating
Not scored
Curated Aggregate·2026-04-0913 workloads / Hopper
VRAM
80 GB
TDP
700 W
Best LLM
No benchmark
MSRP
$25k
Top benchmark
Embedding throughput / 12400 emb/s
The default datacenter inference GPU through 2025. Right answer for production multi-tenant serving. Not for homelab — this is enterprise/cloud territory.
Datacenter GPUNVIDIA
NVIDIA H200 141GB
Rating
Not scored
Curated Aggregate·2026-04-2011 workloads / Hopper
VRAM
141 GB
TDP
700 W
Best LLM
No benchmark
MSRP
$35k
Top benchmark
Embedding throughput / 15200 emb/s
The capacity king. Right answer when 80GB isn't enough and you need 141GB. For everything else, H100 or L40S are more economical.
Datacenter GPUNVIDIA
NVIDIA B200 192GB
Rating
Not scored
Curated Aggregate·2026-02-2212 workloads / Blackwell
VRAM
180 GB
TDP
1000 W
Best LLM
No benchmark
MSRP
$40k
Top benchmark
Embedding throughput / 23500 emb/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUNVIDIA
NVIDIA L40S 48GB
Rating
Not scored
Curated Aggregate·2026-04-299 workloads / Ada Lovelace
VRAM
48 GB
TDP
350 W
Best LLM
No benchmark
MSRP
$7.0k
Top benchmark
Embedding throughput / 10500 emb/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 5090 32GB
Rating
Not scored
Curated Aggregate·2026-08-1030 workloads / Blackwell
VRAM
32 GB
TDP
575 W
Best LLM
No benchmark
MSRP
$2.0k
Top benchmark
Embedding throughput / 12200 emb/s
Consider it for a 32GB CUDA workload after checking the artifact and total system cost. Choose more GPU memory for full-residency 70B Q4.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 4090 24GB
Rating
Not scored
Curated Aggregate·2026-05-2512 workloads / Ada Lovelace
VRAM
24 GB
TDP
450 W
Best LLM
No benchmark
MSRP
$1.6k
Top benchmark
Embedding throughput / 8500 emb/s
Buy if you want the best single-card experience for 32B-class models and either bought at MSRP or found a clean used unit. Skip if 70B is your daily target or if 16GB cards cover your needs.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 4080 Super 16GB
Rating
Not scored
Curated Aggregate·2025-08-0910 workloads / Ada Lovelace
VRAM
16 GB
TDP
320 W
Best LLM
No benchmark
MSRP
$999
Top benchmark
Embedding throughput / 6900 emb/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 4070 Ti 12GB
Rating
Not scored
Curated Aggregate·2025-08-016 workloads / Ada Lovelace
VRAM
12 GB
TDP
285 W
Best LLM
No benchmark
MSRP
$799
Top benchmark
Phi-3 Mini / 165 tok/s
Buy as a gaming-primary, AI-secondary card. For pure AI, the 12GB ceiling is limiting. Consider used 3090 (24GB) or 4060 Ti 16GB instead.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 3090 24GB
Rating
Not scored
Curated Aggregate·2025-05-0110 workloads / Ampere
VRAM
24 GB
TDP
350 W
Best LLM
No benchmark
MSRP
$999
Top benchmark
Embedding throughput / 4900 emb/s
Buy used if you can find a clean unit under $700. The value-for-VRAM proposition is unmatched — nothing else gives you 24GB at this price. Pair with a second used 3090 for 48GB combined VRAM at ~$1,400 total.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 3060 12GB
Rating
Not scored
Curated Aggregate·2025-12-115 workloads / Ampere
VRAM
12 GB
TDP
170 W
Best LLM
No benchmark
MSRP
$329
Top benchmark
Phi-3 Mini / 92.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUAMD
AMD Instinct MI300X 192GB
Rating
Not scored
Curated Aggregate·2026-03-1912 workloads / CDNA 3
VRAM
192 GB
TDP
750 W
Best LLM
No benchmark
MSRP
$15k
Top benchmark
Embedding throughput / 9800 emb/s
Best for organizations with existing AMD infrastructure or those specifically chasing $/VRAM ratios at datacenter scale. The software maturity gap vs NVIDIA is real but narrowing.
Consumer GPUAMD
AMD Radeon RX 7900 XTX 24GB
Rating
Not scored
Curated Aggregate·2026-03-1111 workloads / RDNA 3
VRAM
24 GB
TDP
355 W
Best LLM
No benchmark
MSRP
$899
Top benchmark
Embedding throughput / 5800 emb/s
Buy for Linux AI builds where $/VRAM is the priority and you're comfortable with the ROCm ecosystem. Skip for Windows, for production, or if you value plug-and-play over tinkering.
Your PC result
202.10 tok/s on llama-3.2-3b-instruct
115.60 tok/s on qwen-2.5-7b-instruct / Windows 10 build 26200 + ROCm 7.10 + Ollama 0.24.0 (HIPBLAS). 4 runs per model after a discarded warmup; decode_tps reported as per-run eval_count/eval_duration. Variance <= 0.4% on reproduced models. Captured by site owner on declared rig. Prompt SHA-256: 5e752b688e0761f321a5721d2fa6b57678bf1de1a1180bb4fc9b7e709bafb6a5.
Consumer GPUAMD
AMD Radeon RX 7800 XT 16GB
Rating
Not scored
Curated Aggregate·2025-05-233 workloads / RDNA 3
VRAM
16 GB
TDP
263 W
Best LLM
No benchmark
MSRP
$499
Top benchmark
Phi-3 Mini / 132 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Pro GPUAMD
AMD Radeon Pro W7900 48GB
Rating
Not scored
Curated Aggregate·2025-08-062 workloads / RDNA 3
VRAM
48 GB
TDP
295 W
Best LLM
No benchmark
MSRP
$4.0k
Top benchmark
Llama 3 8B FP16 / 88.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUIntel
Intel Arc B580 12GB
Rating
Not scored
Curated Aggregate·2026-02-157 workloads / Xe2
VRAM
12 GB
TDP
190 W
Best LLM
No benchmark
MSRP
$249
Top benchmark
Phi-3 Mini / 86.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 4070 12GB
Rating
Not scored
Curated Aggregate·2025-08-046 workloads / Ada Lovelace
VRAM
12 GB
TDP
200 W
Best LLM
No benchmark
MSRP
$549
Top benchmark
Phi-3 Mini / 138 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 4060 Ti 16GB
Rating
Not scored
Curated Aggregate·2025-05-016 workloads / Ada Lovelace
VRAM
16 GB
TDP
165 W
Best LLM
No benchmark
MSRP
$499
Top benchmark
Mistral 7B / 56.0 tok/s
Buy for budget AI builds where 16GB VRAM capacity matters more than token generation speed. Best new-card value for local AI experimentation. Skip if you can stretch to a used 3090 at $600-700.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 4060 8GB
Rating
Not scored
Curated Aggregate·2024-09-123 workloads / Ada Lovelace
VRAM
8 GB
TDP
115 W
Best LLM
No benchmark
MSRP
$299
Top benchmark
Phi-3 Mini / 95.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Pro GPUNVIDIA
NVIDIA RTX 6000 Ada 48GB
Rating
Not scored
Curated Aggregate·2025-05-018 workloads / GPU profile
VRAM
48 GB
TDP
300 W
Best LLM
No benchmark
MSRP
$6.8k
Top benchmark
Embedding throughput / 8200 emb/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Pro GPUNVIDIA
NVIDIA A40 48GB
Rating
Not scored
Curated Aggregate·2024-10-114 workloads / GPU profile
VRAM
48 GB
TDP
300 W
Best LLM
No benchmark
MSRP
$5.5k
Top benchmark
Phi-3 Mini / 162 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Pro GPUNVIDIA
NVIDIA RTX A5000 24GB
Rating
Not scored
Curated Aggregate·2024-08-293 workloads / Ampere
VRAM
16 GB
TDP
230 W
Best LLM
No benchmark
MSRP
$1.8k
Top benchmark
Phi-3 Mini / 122 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Pro GPUNVIDIA
NVIDIA RTX A4000 16GB
Rating
Not scored
Curated Aggregate·2024-08-183 workloads / GPU profile
VRAM
16 GB
TDP
140 W
Best LLM
No benchmark
MSRP
$1.2k
Top benchmark
Phi-3 Mini / 92.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUNVIDIA
NVIDIA A100 SXM4 80GB
Rating
Not scored
Curated Aggregate·2025-05-017 workloads / Ampere
VRAM
80 GB
TDP
400 W
Best LLM
No benchmark
MSRP
$15k
Top benchmark
Embedding throughput / 9500 emb/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUNVIDIA
NVIDIA A100 40GB
Rating
Not scored
Curated Aggregate·2024-09-094 workloads / Ampere
VRAM
40 GB
TDP
400 W
Best LLM
No benchmark
MSRP
$10k
Top benchmark
Llama 3 8B FP16 / 195 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUNVIDIA
NVIDIA GH200 Grace Hopper 480GB
Rating
Not scored
Curated Aggregate·2025-02-252 workloads / Hopper
VRAM
576 GB
TDP
1000 W
Best LLM
No benchmark
MSRP
$65k
Top benchmark
Llama 3 8B FP16 / 380 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUAMD
AMD Instinct MI250X 128GB
Rating
Not scored
Curated Aggregate·2024-11-194 workloads / CDNA 2
VRAM
128 GB
TDP
550 W
Best LLM
No benchmark
MSRP
$8.0k
Top benchmark
Llama 3 8B FP16 / 188 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUAMD
AMD Instinct MI210 64GB
Rating
Not scored
Curated Aggregate·2024-09-303 workloads / CDNA 2
VRAM
64 GB
TDP
300 W
Best LLM
No benchmark
MSRP
$17k
Top benchmark
Phi-3 Mini / 175 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUAMD
AMD Radeon RX 7700 XT 12GB
Rating
Not scored
Curated Aggregate·2024-11-212 workloads / RDNA 3
VRAM
12 GB
TDP
245 W
Best LLM
No benchmark
MSRP
$399
Top benchmark
Mistral 7B / 58.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUAMD
AMD Radeon RX 7600 8GB
Rating
Not scored
Curated Aggregate·2024-10-122 workloads / RDNA 3
VRAM
8 GB
TDP
165 W
Best LLM
No benchmark
MSRP
$269
Top benchmark
Phi-3 Mini / 88.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUIntel
Intel Arc A770 16GB
Rating
Not scored
Curated Aggregate·2024-11-154 workloads / GPU profile
VRAM
16 GB
TDP
225 W
Best LLM
No benchmark
MSRP
$349
Top benchmark
Phi-3 Mini / 72.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUIntel
Intel Arc A750 8GB
Rating
Not scored
Curated Aggregate·2024-10-082 workloads / Xe HPG
VRAM
8 GB
TDP
225 W
Best LLM
No benchmark
MSRP
$199
Top benchmark
Phi-3 Mini / 58.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
2× NVIDIA RTX 4090 24GB
Rating
Not scored
Curated Aggregate·2025-03-224 workloads / GPU profile
VRAM
48 GB
TDP
900 W
Best LLM
No benchmark
MSRP
$3.2k
Top benchmark
Gemma 2 9B / 138 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
4× NVIDIA RTX 3090 24GB
Rating
Not scored
Curated Aggregate·2025-01-193 workloads / GPU profile
VRAM
96 GB
TDP
1400 W
Best LLM
No benchmark
MSRP
$4.5k
Top benchmark
Qwen 2.5 14B / 78.0 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUNVIDIA
8× NVIDIA H100 SXM5 80GB
Rating
Not scored
Curated Aggregate·2024-10-044 workloads / GPU profile
VRAM
640 GB
TDP
5600 W
Best LLM
No benchmark
MSRP
$200k
Top benchmark
Llama 3 8B FP16 / 1980 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 5080 16GB
Rating
Not scored
Curated Aggregate·2026-05-2311 workloads / Blackwell
VRAM
16 GB
TDP
360 W
Best LLM
No benchmark
MSRP
$999
Top benchmark
Embedding throughput / 8200 emb/s
Buy for mixed gaming+AI or as a secondary GPU in multi-card rigs. Skip if your primary use case is local AI — used 3090 gives you 24GB at lower cost, and 5070 Ti 16GB gives similar VRAM for $250 less.
Your PC result
429.62 tok/s on kumru-2b
160.83 tok/s on brooqs-mistral-turkish-v2-latest / Batch=1, 2048 context, 512 generated tokens, consecutive warm-model runs. Generation TPS uses Ollama eval_count/eval_duration.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 5070 Ti 16GB
Rating
Not scored
Curated Aggregate·2026-03-048 workloads / Blackwell
VRAM
16 GB
TDP
285 W
Best LLM
No benchmark
MSRP
$749
Top benchmark
Embedding throughput / 7100 emb/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 5070 12GB
Rating
Not scored
Curated Aggregate·2026-03-206 workloads / Blackwell
VRAM
12 GB
TDP
250 W
Best LLM
No benchmark
MSRP
$549
Top benchmark
Phi-3 Mini / 165 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 5060 Ti 16GB
Rating
Not scored
Curated Aggregate·2026-04-225 workloads / Blackwell
VRAM
16 GB
TDP
180 W
Best LLM
No benchmark
MSRP
$429
Top benchmark
Phi-3 Mini / 132 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 5060 Ti 8GB
Rating
Not scored
Curated Aggregate·2026-04-162 workloads / Blackwell
VRAM
8 GB
TDP
180 W
Best LLM
No benchmark
MSRP
$379
Top benchmark
Phi-3 Mini / 158 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Consumer GPUNVIDIA
NVIDIA GeForce RTX 5060 8GB
Rating
Not scored
Curated Aggregate·2026-04-283 workloads / Blackwell
VRAM
8 GB
TDP
150 W
Best LLM
No benchmark
MSRP
$299
Top benchmark
Phi-3 Mini / 145 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUNVIDIA
NVIDIA DGX Spark (Project DIGITS, 128GB)
Rating
Not scored
Curated Aggregate·2026-05-258 workloads / GPU profile
VRAM
128 GB
TDP
240 W
Best LLM
No benchmark
MSRP
$3.0k
Top benchmark
DeepSeek-R1 7B / 145 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUAMD
AMD Instinct MI300X 192GB
Rating
Not scored
Curated Aggregate·2025-04-224 workloads / GPU profile
VRAM
192 GB
TDP
750 W
Best LLM
No benchmark
MSRP
$15k
Top benchmark
Embedding throughput / 14800 emb/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.
Datacenter GPUNVIDIA
NVIDIA GH200 (144GB HBM configuration)
Rating
Not scored
Curated Aggregate·2024-10-262 workloads / GPU profile
VRAM
144 GB
TDP
1000 W
Best LLM
No benchmark
MSRP
$45k
Top benchmark
Mistral 7B / 305 tok/s
Open the hardware brief for methodology, confidence ranges, and buyer guidance.