Hardware / NVIDIA /L40S

NVIDIA · GPU

L40S

Runs 134 models fully in memory, 14 more with CPU offload.

Memory

48 GB

Bandwidth

864 GB/s

Runs

134

01

Specifications

Memory

48 GB

Decides which models fit at all.

Memory bandwidth

864 GB/s

How fast it can read the model. Sets the speed of the reply.

Compute

91.6 TFLOPS

Trillions of half-precision sums a second. Sets how fast it reads a long prompt, not how fast it replies.

Power draw

350 W

At full load, for sizing a power supply.

MSRP

$7,500

What the maker asked for it new. We hold no used or street prices, so this is not what one costs today.

Released

Aug 8, 2023

Architecture

Ada Lovelace

02

What it runs

Best quant per model. Text models first, then speech, then what does not fit.

What is quantisation? →

ModelParamsContextSpeedVerdict

Granite 4.1 8B8.8B131K130 tok/sFits in memory Qwen3.6 35B A3B36B131K275 tok/sFits in memory Gemma 4 26B A4B25.8B66K207 tok/sFits in memory Nemotron 3 Nano 30B A3B31.6B262K275 tok/sFits in memory Ministral 3 8B 25128.9B262K128 tok/sFits in memory Ministral 3 3B 25123.8B131K171 tok/sFits in memory gpt-oss-safeguard-20b21.5B131K269 tok/sFits in memory Granite 4.0 Micro3.2B66K304 tok/sFits in memory Qwen3 VL 8B Thinking8.8B131K130 tok/sFits in memory Qwen3 VL 30B A3B Thinking31.1B131K323 tok/sFits in memory UI-TARS 7B8.3B66K137 tok/sFits in memory Llama 3.2 3B Instruct3.2B131K304 tok/sFits in memory Qwen3.5-35B-A3B36B262K323 tok/sFits in memory Qwen3 VL 8B Instruct8.8B262K130 tok/sFits in memory Qwen3 VL 30B A3B Instruct31.1B131K275 tok/sFits in memory Qwen3 30B A3B Thinking 250730.5B66K184 tok/sFits in memory Qwen3 Coder 30B A3B Instruct30.5B66K184 tok/sFits in memory Qwen3 30B A3B Instruct 250730.5B262K323 tok/sFits in memory Qwen3 8B8.2B131K139 tok/sFits in memory Gemma 3 4B4.3B131K226 tok/sFits in memory Llama 3.1 8B Instruct8B131K143 tok/sFits in memory Granite 4.2 8B8.8B131K130 tok/sFits in memory Nemotron 3.5 Content Safety4.3B131K226 tok/sFits in memory Qwen3.5-9B9.7B262K118 tok/sFits in memory Gemma 3n 4B7.8B33K125 tok/sFits in memory Llama 3.2 1B Instruct1.2B33K810 tok/sFits in memory Qwen2.5 7B Instruct7.6B33K128 tok/sFits in memory Reka Edge7.1B16K137 tok/sFits in memory Llama Guard 4 12B12B262K95 tok/sFits in memory Gemma 3 12B12.2B131K94 tok/sFits in memory Rocinante 12B12.2B66K94 tok/sFits in memory Mistral Nemo12.2B131K94 tok/sFits in memory Ministral 3 14B 251213.9B131K82 tok/sFits in memory Llama 3 8B Lunaris8B8K122 tok/sFits in memory Hy-MT2-1.8B2B8K485 tok/sFits in memory Hy-MT2-30B-A3B30.1B8K275 tok/sFits in memory Hy-MT2-7B8B8K121 tok/sFits in memory UnslopNemo 12B12.2B33K94 tok/sFits in memory Qwen3 14B14.8B131K77 tok/sFits in memory Reka Flash 320.9B66K55 tok/sFits in memory Phi 414.7B16K78 tok/sFits in memory MythoMax 13B13B8K88 tok/sFits in memory Cydonia 24B V4.123.6B131K48 tok/sFits in memory Uncensored24B66K48 tok/sFits in memory Mistral Small 3.2 24B24B131K48 tok/sFits in memory Mistral Small 3.1 24B24B66K48 tok/sFits in memory Ternary Bonsai 2 27B27B262K42 tok/sFits in memory Qwen3.5-27B27.8B66K41 tok/sFits in memory Gemma 3 27B27.4B262K42 tok/sFits in memory ReMM SLERP 13B13B4K88 tok/sFits in memory Qwen3.6 27B27.8B66K41 tok/sFits in memory Muse Glimmer 30B29.8B131K38 tok/sFits in memory Qwen3.8 27B27.8B66K41 tok/sFits in memory Olmo 3 32B Think32.2B66K35 tok/sFits in memory Qwen3 VL 32B Instruct33.4B66K34 tok/sFits in memory Qwen3 32B32.8B66K35 tok/sFits in memory Mistral Small 323.6B33K48 tok/sFits in memory GLM 4.7 Flash31.2B66K31 tok/sFits in memory Nemotron 3.5 Lightning31.6B262K31 tok/sFits in memory Nex-N2-Mini35.1B262K28 tok/sFits in memory Laguna XS 2.133.4B131K29 tok/sFits in memory Nex-N2.5-Mini35.1B262K28 tok/sFits in memory Voxtral Small 24B 250724.3B16K47 tok/sFits in memory Qwen2.5 Coder 32B Instruct32.8B33K35 tok/sFits in memory Skyfall 36B V236.9B33K31 tok/sFits in memory Gemma 4 31B31.3B16K36 tok/sFits in memory Gemma 2 27B27.2B8K42 tok/sFits in memory

Speech models it runs67 modelsTranscription and speech models that fit in memory. Throughput figures for a transcription model come from its text decoder, so treat this as a fit answer rather than a speed one.

Niagara 19m Batch.en20M32527 tok/sFits in memory Niagara 38m Batch.en38M17120 tok/sFits in memory Higgs Audio v3 8b STT v28.9B128 tok/sFits in memory Higgs Audio v3 STT2.7B241 tok/sFits in memory Cohere Transcribe 03 20262.1B310 tok/sFits in memory Distil Large v3.50.8B813 tok/sFits in memory Lite Whisper Large v3 Acc1.4B694 tok/sFits in memory Owsm CTC v3.1 1B1.1B1037 tok/sFits in memory Owsm CTC v3.2 ft 1B1B651 tok/sFits in memory Owsm CTC v4 1B1B1140 tok/sFits in memory Hubert Large Ls960 ft0.3B2033 tok/sFits in memory MMS 1b All1B972 tok/sFits in memory Wav2vec2 Large 960h Lv60 Self0.3B2033 tok/sFits in memory Hojo ASR V15.2B188 tok/sFits in memory Granite 4.0 1b Speech2.3B283 tok/sFits in memory Granite Speech 3.3 2b3B380 tok/sFits in memory Granite Speech 3.3 8b8.6B133 tok/sFits in memory Granite Speech 4.1 2b2.3B496 tok/sFits in memory Granite Speech 4.1 2b NAR2.3B496 tok/sFits in memory STT 2.6b en2.6B439 tok/sFits in memory Phi 4 Multimodal Instruct5.6B131K204 tok/sFits in memory VibeVoice ASR HF8.3B137 tok/sFits in memory Voxtral Mini 3B 25074.7B138 tok/sFits in memory Voxtral Mini 4B Realtime 26024.4B148 tok/sFits in memory Canary 180m Flash0.2B5399 tok/sFits in memory Canary 1b1B651 tok/sFits in memory Canary 1b Flash0.8B813 tok/sFits in memory Canary 1b v21B651 tok/sFits in memory Canary Qwen 2.5b2.6B250 tok/sFits in memory Nemotron 3.5 ASR Streaming 0.6b0.6B1084 tok/sFits in memory Nemotron Speech Streaming en 0.6b0.6B1084 tok/sFits in memory Parakeet CTC 0.6b0.6B1900 tok/sFits in memory Parakeet CTC 1.1b1.1B591 tok/sFits in memory Parakeet RNNT 0.6b0.6B1084 tok/sFits in memory Parakeet RNNT 1.1b1.1B1037 tok/sFits in memory Parakeet TDT CTC 110m0.1B5914 tok/sFits in memory Parakeet TDT 0.6b v20.6B1620 tok/sFits in memory Parakeet TDT 0.6b v30.6B1084 tok/sFits in memory Parakeet TDT 1.1b1.1B591 tok/sFits in memory Nyra Health CrisperWhisper1.6B407 tok/sFits in memory Whisper Large v31.5B648 tok/sFits in memory Whisper Large v3 Turbo0.8B1425 tok/sFits in memory MOSS Transcribe Diarize0.9B723 tok/sFits in memory Qwen3 ASR 0.6B0.9B1267 tok/sFits in memory Qwen3 ASR 0.6B HF0.8B1425 tok/sFits in memory Qwen3 ASR 1.7B2.3B423 tok/sFits in memory Qwen3 ASR 1.7B HF2B325 tok/sFits in memory Zipformer cr CTC Transducer XL 290M0.3B3351 tok/sFits in memory Zipformer Transducer XL 290M0.3B2243 tok/sFits in memory ASR Conformer Largescaleasr0.5B2375 tok/sFits in memory GLM ASR Nano 25122.3B423 tok/sFits in memory Granite Speech 5.0 470m Turboctc0.5B1944 tok/sFits in memory Granite Speech 5.0 470m Turboctc nc0.5B2280 tok/sFits in memory Fun ASR Nano 2512 HF0.8B1425 tok/sFits in memory Gemma 4 E2B it5.1B128 tok/sFits in memory Gemma 4 E4B it8B131K122 tok/sFits in memory Niagara 9m Batch.en9M72282 tok/sFits in memory Niagara 84m Batch.en84M13573 tok/sFits in memory ARK ASR 0.6B1.3B33K748 tok/sFits in memory ARK ASR 3B4.1B33K278 tok/sFits in memory Audio8 ASR 0.1B0.3B33K2169 tok/sFits in memory MOSS Transcribe Preview 2B2.4B33K475 tok/sFits in memory Gemma 4 12B it12B95 tok/sFits in memory Moonshine Streaming Medium0.3B4K3800 tok/sFits in memory Moonshine Streaming Small0.1B4K6505 tok/sFits in memory Moonshine Streaming Tiny0B4Kno speed estimateFits in memory Moonshine Tiny0B194no speed estimateFits in memory

Runs only by spilling into system memory14 modelsThese load, but part of the weights sits in ordinary system memory, which is far slower than the chip.

Qwen3 Coder Next79.7B—no speed estimateSpills to system RAM Qwen3 Next 80B A3B Thinking81.3B—no speed estimateSpills to system RAM Hermes 4 70B70.6B—no speed estimateSpills to system RAM Hunyuan A13B Instruct80.4B—no speed estimateSpills to system RAM Qwen2.5 VL 72B Instruct73.4B—no speed estimateSpills to system RAMest Llama 3.3 Euryale 70B70.6B—no speed estimateSpills to system RAM Magnum v4 72B72.7B—no speed estimateSpills to system RAM Llama 3.1 Euryale 70B v2.270.6B—no speed estimateSpills to system RAMest Qwen3 Next 80B A3B Instruct81.3B—no speed estimateSpills to system RAM R1 Distill Llama 70B70.6B—no speed estimateSpills to system RAMest Qwen2.5 72B Instruct72.7B—no speed estimateSpills to system RAMest Hermes 3 70B Instruct70.6B—no speed estimateSpills to system RAM Llama 3.1 70B Instruct70.6B—no speed estimateSpills to system RAMest Llama 3.3 70B Instruct70.6B—no speed estimateSpills to system RAMest

Too large for this device76 modelsThe weights do not fit even with part of them offloaded to system memory. Closest calls first.

Nemotron 3 Super124B—no speed estimateToo large GLM 4.5V108B—no speed estimateToo large Llama 4 Scout109B—no speed estimateToo large GLM 4.5 Air111B—no speed estimateToo large Command A111B—no speed estimateToo large Laguna S 2.1118B—no speed estimateToo large gpt-oss-120b120B—no speed estimateToo large Mistral Small 4119B—no speed estimateToo large Devstral 2 2512125B—no speed estimateToo large Mixtral 8x22B Instruct141B—no speed estimateToo large GLM 4.6V108B—no speed estimateToo large Ling-2.6-flash107B—no speed estimateToo large Step 3.7 Flash201B—no speed estimateToo large Qwen3.5-122B-A10B125B—no speed estimateToo large Mistral Medium 3.5128B—no speed estimateToo large Ling-3.0-flash128B—no speed estimateToo large MiniMax M2.7229B—no speed estimateToo large Step 3.5 Flash199B—no speed estimateToo large Qwen3 VL 235B A22B Thinking236B—no speed estimateToo large MiniMax M2.5229B—no speed estimateToo large DeepSeek V4 Flash291B—no speed estimateToo large GLM 5.3 Flash321B—no speed estimateToo large Hy3299B—no speed estimateToo large GLM 4.5358B—no speed estimateToo large Laguna M.1226B—no speed estimateToo large Nex-N2-Pro397B—no speed estimateToo large Nex-N2.5-Pro397B—no speed estimateToo large MiniMax M2.1229B—no speed estimateToo large MiniMax M2229B—no speed estimateToo large Qwen3 235B A22B Thinking 2507235B—no speed estimateToo large Qwen3 235B A22B Instruct 2507235B—no speed estimateToo large Qwen3 VL 235B A22B Instruct236B—no speed estimateToo large GLM 4.6357B—no speed estimateToo large GLM 4.7358B—no speed estimateToo large Inkling Small266B—no speed estimateToo large Trinity Large Thinking399B—no speed estimateToo large Jamba Large 1.7399B—no speed estimateToo large Hermes 4 405B406B—no speed estimateToo large MiniMax M3427B—no speed estimateToo large Hy3 preview299B—no speed estimateToo large DeepSeek V4 Flash Vision Exp305B—no speed estimateToo large MiniMax M1456B—no speed estimateToo large MiMo-V2.5311B—no speed estimateToo large Qwen3 Coder 480B A35B480B—no speed estimateToo large Nemotron 3 Ultra561B—no speed estimateToo large Llama 4 Maverick402B—no speed estimateToo large Qwen3.5 397B A17B403B—no speed estimateToo large Hermes 3 405B Instruct406B—no speed estimateToo large ERNIE 4.5 VL 424B A47B424B—no speed estimateToo large GLM 5.1754B—no speed estimateToo large GLM 5.2753B—no speed estimateToo large MiniMax-01456B—no speed estimateToo large R1 0528685B—no speed estimateToo large DeepSeek V3685B—no speed estimateToo large DeepSeek V3.1685B—no speed estimateToo large GLM 5754B—no speed estimateToo large DeepSeek V4.1 Flash763B—no speed estimateToo large MiMo-V2.5-Pro1T—no speed estimateToo large Kimi K2 07111T—no speed estimateToo large Ring-2.6-1T1T—no speed estimateToo large Ling-2.6-1T1T—no speed estimateToo large DeepSeek V3.1 Terminus685B—no speed estimateToo large DeepSeek V3 0324685B—no speed estimateToo large DeepSeek V3.2685B—no speed estimateToo large DeepSeek V3.2 Exp685B—no speed estimateToo large Kimi K2.7 Code1.1T—no speed estimateToo large Kimi K2.61.1T—no speed estimateToo large Kimi K2.51.1T—no speed estimateToo large Hy4 preview780B—no speed estimateToo large Inkling952B—no speed estimateToo large Kimi K2 09051T—no speed estimateToo large Kimi K2 Thinking1.1T—no speed estimateToo large Qwen3.8 2.4T A95B2.4T—no speed estimateToo large Kimi K32.8T—no speed estimateToo large DeepSeek V4 Pro1.6T—no speed estimateToo large LongCat 2.01.8T—no speed estimateToo large

224 models have a verdict on this device: 134 fit in memory, 14 spill into system memory and 76 are too large. Every one of them is on this page.

03

Other Ada Lovelace cards

The same generation in other memory sizes. Memory is what decides which models fit.

GeForce RTX 4060 8GB8 GB · less memory GeForce RTX 4070 SUPER12 GB · less memory GeForce RTX 4060 Ti 16GB16 GB · less memory GeForce RTX 4070 Ti SUPER16 GB · less memory GeForce RTX 4080 SUPER16 GB · less memory GeForce RTX 409024 GB · less memory RTX 6000 Ada48 GB · same memory

Something wrong on this page? Tell us