Compare › RTX 5090 vs RTX 5080

RTX 5090 vs RTX 5080 in Australia

Prices checked 21 Oct 2026, 10:07 am AEST · how we check

Today the cheapest RTX 5090 we found is A$10,999 at Scorptec (in stock) and the RTX 5080 is A$2,323 at Mwave (in stock). The RTX 5080 costs A$8,676 less (79%). For local AI, the RTX 5090 gives models 32 GB at 1,792 GB/s and the RTX 5080 16 GB at 960 GB/s, so the RTX 5090 generates text about 1.6× faster on models both can hold. The RTX 5090 runs 10 models we track that the RTX 5080 can't at 4-bit.

Side by side

Lowest AU price today
RTX 5090A$10,999
Scorptec · in stock
RTX 5080A$2,323
Mwave · in stock
Memory for models
RTX 509032 GB
of 32 GB VRAM
RTX 508016 GB
of 16 GB VRAM
Memory bandwidth
RTX 50901,792 GB/s
RTX 5080960 GB/s
Speed (Llama 2 7B Q4_0)
RTX 5090290 tok/s measured
RTX 5080182 tok/s measured
Speed per A$1,000
RTX 509026.4 tok/s
RTX 508078.3 tok/s
Models that fit (4-bit, 8k)
RTX 509020 of 32
RTX 508010 of 32
Biggest model that fits

Which new models each one runs

The newest open models we track that at least one of them can hold, at the usual 4-bit quant with an 8k context. Speeds are bandwidth-based estimates. 6 of the 10 newest models are too big for both.

Hy Memory 30B A3B (MoE) new (Q4_K_M, needs 20 GB)
RTX 5090fits ~605 tok/s
RTX 5080doesn't fit
LFM2.5 24B-A2B (MoE) new (Q4_K_M, needs 16 GB)
RTX 5090fits ~907 tok/s
RTX 5080doesn't fit
MiniCPM V 4.7 35B A3B (MoE) new (Q4_K_M, needs 23 GB)
RTX 5090fits ~605 tok/s
RTX 5080doesn't fit
MiniCPM V 4.7 1B new (Q4_K_M, needs 2 GB)
RTX 5090fits ~1170 tok/s
RTX 5080fits ~734 tok/s
Qwen3.8 27B (Q4_K_M, needs 18 GB)
RTX 5090fits ~65 tok/s
RTX 5080doesn't fit
Nemotron 3.5 Lightning 30B-A3B (MoE) (Q4_K_M, needs 20 GB)
RTX 5090fits ~605 tok/s
RTX 5080doesn't fit
Gemma 4 12B (Q4_K_M, needs 9 GB)
RTX 5090fits ~152 tok/s
RTX 5080fits ~95 tok/s
Qwen3.6 35B-A3B (MoE) (Q4_K_M, needs 23 GB)
RTX 5090fits ~605 tok/s
RTX 5080doesn't fit

FAQ

Which is cheaper in Australia, the RTX 5090 or the RTX 5080?

The RTX 5080: A$2,323 at Mwave against A$10,999 for the RTX 5090 (Scorptec), checked 21 Oct 2026, 10:07 am AEST. We re-check both every day.

Is the RTX 5090 faster than the RTX 5080 for local AI?

Yes. On models that fit in both, text generation speed mostly follows memory bandwidth: 1,792 GB/s vs 960 GB/s. Our reference test gives 290 vs 182 tok/s (measured / measured).

Which runs bigger AI models?

The RTX 5090, with 32 GB usable vs 16 GB. Models only it can hold at 4-bit include Hy Memory 30B A3B (MoE), LFM2.5 24B-A2B (MoE), MiniCPM V 4.7 35B A3B (MoE).

Stores change prices during the day, so check the listing before you pay. How we check

Related

RTX 5090 price in AustraliaRTX 5080 price in AustraliaMac Studio vs RTX 5090RTX PRO 6000 vs RTX 5090RTX 5080 vs RTX 5070 TiDGX Spark alternativesRTX 5090 in stockAll comparisonsAll AU prices

Weekly email

Monday mornings: which open models came out, how much memory they need, and which AU prices moved. Unsubscribe any time.

Used for this email only. Privacy