· Support

Production support.

Every configuration listed here is production ready and tested on the named hardware.

Coverage grid

Every model, every architecture, at a glance.

Columns are every chip with production support on the current release. A green dot means the model runs there in production.

Production on the current release Not supported

102 of 102 configurations

102 configurations match

Install Photon and run locally.

$pip install moondream
PrecisionCapabilitiesBenchmark
Gemma 4 E2B GemmaNVIDIA AmpereNVIDIA A10/A10GBF16Query · Captionv2.2Sep 9, 2026–
Gemma 4 E2B GemmaNVIDIA AmpereNVIDIA A100 SXM 80GBBF16Query · Captionv2.2Sep 9, 2026–
Gemma 4 E2B GemmaNVIDIA BlackwellNVIDIA B200BF16Query · Captionv2.1Aug 30, 2026Results
Gemma 4 E2B GemmaNVIDIA HopperH100 SXM 80GBBF16Query · Captionv2.0Aug 3, 2026Results
Gemma 4 E2B GemmaNVIDIA Ada LovelaceNVIDIA L4BF16Query · Captionv2.2Sep 9, 2026–
Gemma 4 E2B GemmaNVIDIA AmpereNVIDIA RTX 3090BF16Query · Captionv2.2Sep 9, 2026–
Gemma 4 E2B GemmaNVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Query · Captionv2.2Sep 9, 2026–
Gemma 4 E4B GemmaNVIDIA AmpereNVIDIA A10/A10GBF16Query · Captionv2.2Sep 9, 2026–
Gemma 4 E4B GemmaNVIDIA AmpereNVIDIA A100 SXM 80GBBF16Query · Captionv2.2Sep 9, 2026–
Gemma 4 E4B GemmaNVIDIA BlackwellNVIDIA B200BF16Query · Captionv2.1Aug 30, 2026Results
Gemma 4 E4B GemmaNVIDIA HopperH100 SXM 80GBBF16Query · Captionv2.0Aug 3, 2026Results
Gemma 4 E4B GemmaNVIDIA Ada LovelaceNVIDIA L4BF16Query · Captionv2.2Sep 9, 2026–
Gemma 4 E4B GemmaNVIDIA AmpereNVIDIA RTX 3090BF16Query · Captionv2.2Sep 9, 2026–
Gemma 4 E4B GemmaNVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Query · Captionv2.2Sep 9, 2026–
Moondream 2 Moondream VLMNVIDIA AmpereA10 24GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026Results
Moondream 2 Moondream VLMNVIDIA AmpereA40 48GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026–
Moondream 2 Moondream VLMNVIDIA AmpereA100 80GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026Results
Moondream 2 Moondream VLMNVIDIA BlackwellNVIDIA B200BF16Query · Caption · Detect · Point · Segmentv2.1Aug 31, 2026–
Moondream 2 Moondream VLMNVIDIA HopperGH200 96GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026–
Moondream 2 Moondream VLMNVIDIA HopperH100 SXM 80GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 3, 2026Results
Moondream 2 Moondream VLMNVIDIA HopperH200 SXM 141GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026–
Moondream 2 Moondream VLMNVIDIA AmpereJetson AGX Orin 64GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026Results
Moondream 2 Moondream VLMNVIDIA AmpereJetson Orin Nano 8GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026–
Moondream 2 Moondream VLMNVIDIA AmpereJetson Orin NX 16GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026–
Moondream 2 Moondream VLMNVIDIA Ada LovelaceL4 24GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026Results
Moondream 2 Moondream VLMNVIDIA Ada LovelaceL40S 48GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026Results
Moondream 2 Moondream VLMNVIDIA AmpereRTX 3090 24GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026–
Moondream 2 Moondream VLMNVIDIA AmpereNVIDIA RTX 3090BF16Query · Caption · Detect · Point · Segmentv2.2Sep 9, 2026–
Moondream 2 Moondream VLMNVIDIA Ada LovelaceRTX 4090 24GBBF16Query · Caption · Detect · Point · Segmentv1.0Aug 4, 2026–
Moondream 2 Moondream VLMNVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Query · Caption · Detect · Point · Segmentv2.2Sep 9, 2026–
Moondream 3 Moondream VLMNVIDIA AmpereA10 24GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026Results
Moondream 3 Moondream VLMNVIDIA AmpereA40 48GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026–
Moondream 3 Moondream VLMNVIDIA AmpereA100 80GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026Results
Moondream 3 Moondream VLMNVIDIA BlackwellNVIDIA B200BF16Query · Caption · Detect · Point · Segmentv2.1Aug 30, 2026Results
Moondream 3 Moondream VLMNVIDIA HopperGH200 96GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026–
Moondream 3 Moondream VLMNVIDIA HopperH100 SXM 80GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 3, 2026Results
Moondream 3 Moondream VLMNVIDIA HopperH200 SXM 141GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026–
Moondream 3 Moondream VLMNVIDIA AmpereJetson AGX Orin 64GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026Results
Moondream 3 Moondream VLMNVIDIA AmpereJetson Orin Nano 8GBBF16Query · Caption · Detect · Point · Segmentv2.1Sep 1, 2026–
Moondream 3 Moondream VLMNVIDIA AmpereJetson Orin NX 16GBBF16Query · Caption · Detect · Point · Segmentv2.1Sep 1, 2026–
Moondream 3 Moondream VLMNVIDIA Ada LovelaceL4 24GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026Results
Moondream 3 Moondream VLMNVIDIA Ada LovelaceL40S 48GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026Results
Moondream 3 Moondream VLMNVIDIA AmpereRTX 3090 24GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026–
Moondream 3 Moondream VLMNVIDIA AmpereNVIDIA RTX 3090BF16Query · Caption · Detect · Point · Segmentv2.2Sep 9, 2026–
Moondream 3 Moondream VLMNVIDIA Ada LovelaceRTX 4090 24GBBF16Query · Caption · Detect · Point · Segmentv2.0Aug 4, 2026–
Moondream 3 Moondream VLMNVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Query · Caption · Detect · Point · Segmentv2.2Sep 9, 2026–
Parakeet TDT 0.6B v3 ParakeetNVIDIA AmpereNVIDIA A10/A10GBF16Transcribev2.2Sep 9, 2026Results
Parakeet TDT 0.6B v3 ParakeetNVIDIA AmpereNVIDIA A100 SXM 80GBBF16Transcribev2.2Sep 9, 2026Results
Parakeet TDT 0.6B v3 ParakeetNVIDIA BlackwellNVIDIA B200BF16Transcribev2.1Aug 31, 2026Results
Parakeet TDT 0.6B v3 ParakeetNVIDIA HopperH100 SXM 80GBBF16Transcribev2.1Aug 31, 2026Results
Parakeet TDT 0.6B v3 ParakeetNVIDIA Ada LovelaceNVIDIA L4BF16Transcribev2.2Sep 9, 2026Results
Parakeet TDT 0.6B v3 ParakeetNVIDIA AmpereNVIDIA RTX 3090BF16Transcribev2.2Sep 9, 2026–
Parakeet TDT 0.6B v3 ParakeetNVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Transcribev2.2Sep 9, 2026–
Qwen 3.5 0.8B Qwen 3NVIDIA AmpereNVIDIA A10/A10GBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 0.8B Qwen 3NVIDIA AmpereNVIDIA A100 SXM 80GBBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 0.8B Qwen 3NVIDIA BlackwellNVIDIA B200BF16Query · Captionv2.1Aug 30, 2026Results
Qwen 3.5 0.8B Qwen 3NVIDIA HopperH100 SXM 80GBBF16Query · Captionv2.0Aug 3, 2026Results
Qwen 3.5 0.8B Qwen 3NVIDIA Ada LovelaceNVIDIA L4BF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 0.8B Qwen 3NVIDIA AmpereNVIDIA RTX 3090BF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 0.8B Qwen 3NVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 2B Qwen 3NVIDIA AmpereNVIDIA A10/A10GBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 2B Qwen 3NVIDIA AmpereNVIDIA A100 SXM 80GBBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 2B Qwen 3NVIDIA BlackwellNVIDIA B200BF16Query · Captionv2.1Aug 30, 2026Results
Qwen 3.5 2B Qwen 3NVIDIA HopperH100 SXM 80GBBF16Query · Captionv2.0Aug 3, 2026Results
Qwen 3.5 2B Qwen 3NVIDIA Ada LovelaceNVIDIA L4BF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 2B Qwen 3NVIDIA AmpereNVIDIA RTX 3090BF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 2B Qwen 3NVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 4B Qwen 3NVIDIA AmpereNVIDIA A10/A10GBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 4B Qwen 3NVIDIA AmpereNVIDIA A100 SXM 80GBBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 4B Qwen 3NVIDIA BlackwellNVIDIA B200BF16Query · Captionv2.1Aug 30, 2026Results
Qwen 3.5 4B Qwen 3NVIDIA HopperH100 SXM 80GBBF16Query · Captionv2.0Aug 3, 2026Results
Qwen 3.5 4B Qwen 3NVIDIA Ada LovelaceNVIDIA L4BF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 4B Qwen 3NVIDIA AmpereNVIDIA RTX 3090BF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 4B Qwen 3NVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 9B Qwen 3NVIDIA AmpereNVIDIA A10/A10GBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 9B Qwen 3NVIDIA AmpereNVIDIA A100 SXM 80GBBF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 9B Qwen 3NVIDIA BlackwellNVIDIA B200BF16Query · Captionv2.1Aug 30, 2026Results
Qwen 3.5 9B Qwen 3NVIDIA HopperH100 SXM 80GBBF16Query · Captionv2.0Aug 3, 2026Results
Qwen 3.5 9B Qwen 3NVIDIA Ada LovelaceNVIDIA L4BF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 9B Qwen 3NVIDIA AmpereNVIDIA RTX 3090BF16Query · Captionv2.2Sep 9, 2026–
Qwen 3.5 9B Qwen 3NVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Query · Captionv2.2Sep 9, 2026–
Qwen3-ASR 0.6B Qwen3-ASRNVIDIA AmpereNVIDIA A10/A10GBF16Transcribev2.2Sep 9, 2026Results
Qwen3-ASR 0.6B Qwen3-ASRNVIDIA AmpereNVIDIA A100 SXM 80GBBF16Transcribev2.2Sep 9, 2026Results
Qwen3-ASR 0.6B Qwen3-ASRNVIDIA BlackwellNVIDIA B200BF16Transcribev2.1Aug 31, 2026Results
Qwen3-ASR 0.6B Qwen3-ASRNVIDIA HopperH100 SXM 80GBBF16Transcribev2.1Aug 31, 2026Results
Qwen3-ASR 0.6B Qwen3-ASRNVIDIA Ada LovelaceNVIDIA L4BF16Transcribev2.2Sep 9, 2026Results
Qwen3-ASR 0.6B Qwen3-ASRNVIDIA AmpereNVIDIA RTX 3090BF16Transcribev2.2Sep 9, 2026–
Qwen3-ASR 0.6B Qwen3-ASRNVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Transcribev2.2Sep 9, 2026–
Qwen3-ASR 1.7B Qwen3-ASRNVIDIA AmpereNVIDIA A10/A10GBF16Transcribev2.2Sep 9, 2026Results
Qwen3-ASR 1.7B Qwen3-ASRNVIDIA AmpereNVIDIA A100 SXM 80GBBF16Transcribev2.2Sep 9, 2026Results
Qwen3-ASR 1.7B Qwen3-ASRNVIDIA BlackwellNVIDIA B200BF16Transcribev2.1Aug 31, 2026Results
Qwen3-ASR 1.7B Qwen3-ASRNVIDIA HopperH100 SXM 80GBBF16Transcribev2.1Aug 31, 2026Results
Qwen3-ASR 1.7B Qwen3-ASRNVIDIA Ada LovelaceNVIDIA L4BF16Transcribev2.2Sep 9, 2026Results
Qwen3-ASR 1.7B Qwen3-ASRNVIDIA AmpereNVIDIA RTX 3090BF16Transcribev2.2Sep 9, 2026–
Qwen3-ASR 1.7B Qwen3-ASRNVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Transcribev2.2Sep 9, 2026–
Whisper large-v3-turbo WhisperNVIDIA AmpereNVIDIA A10/A10GBF16Transcribev2.2Sep 9, 2026–
Whisper large-v3-turbo WhisperNVIDIA AmpereNVIDIA A100 SXM 80GBBF16Transcribev2.2Sep 9, 2026–
Whisper large-v3-turbo WhisperNVIDIA BlackwellNVIDIA B200BF16Transcribev2.1Aug 31, 2026Results
Whisper large-v3-turbo WhisperNVIDIA HopperH100 SXM 80GBBF16Transcribev2.1Aug 31, 2026Results
Whisper large-v3-turbo WhisperNVIDIA Ada LovelaceNVIDIA L4BF16Transcribev2.2Sep 9, 2026–
Whisper large-v3-turbo WhisperNVIDIA AmpereNVIDIA RTX 3090BF16Transcribev2.2Sep 9, 2026–
Whisper large-v3-turbo WhisperNVIDIA BlackwellNVIDIA RTX PRO 6000 BlackwellBF16Transcribev2.2Sep 9, 2026–

Updated by release · updates ·