Mochi 1 Preview · 3 GPUs measured first-party · text-to-video · Updated October 2026
What GPU Do You Need for Mochi 1 Preview?
Mochi 1 Preview on 3 GPUs, measured first-party: NVIDIA H200 leads at 0.4 frames/s, A100 80GB SXM4 trails at 0.2 frames/s, and it peaked at 71GB of VRAM.
Benchmarked weights: genmo/mochi-1-preview
NVIDIA H200
0.4 frames/s on Mochi 1 Preview, the ceiling. Measured on our bench. 141GB of VRAM, $31,000 at launch.
- 0.4 frames/s on Mochi 1 Preview
- 141GB, clears the Mochi 1 Preview floor
- Rentable by the hour rather than bought
- 700W board rating
- Datacenter or workstation hardware, not a retail purchase
What GPU Do You Need for Mochi 1 Preview?, frames/s by GPU
Efficiency: frames/s per 100W drawn
Power is the average pulled during the run, sampled at 1Hz. The fastest card is often not the one here, and for anything left running this is the number that shows up on the bill.
Value: frames/s per $1,000 of MSRP
Launch price, not street price, so it ages. A speed leaderboard always crowns the most expensive card; this is the counterweight.
Mochi 1 Preview. Measured text-to-video speed by GPU
| GPU | frames/s | Peak VRAM | Avg power |
|---|---|---|---|
| NVIDIA H200 | 0.4 | 67.3GB | 682.6 W |
| NVIDIA H100 80GB HBM3 | 0.38 | 67.1GB | 683.2 W |
| NVIDIA A100 80GB SXM4 | 0.2 | 67.2GB | 373.5 W |
What the numbers show. Across 3 GPUs measured on our own bench, H200 is fastest at 0.40 frames/s. The slowest, A100 80GB SXM4, manages 0.20, so the spread is 2.0x from top to bottom.
How it compares on the same H100. Mochi took 128 seconds for a 1.6-second clip at 848x480, about 79 GPU-seconds per second of video. On the same card Wan 2.2 T2V A14B took 146 seconds for a 3.1-second clip at 832x480 (about 48 per second of video), CogVideoX-2B 42 seconds for a 6.1-second clip at 720x480 (about 7), and LTX-Video under 6 seconds per clip. So Mochi is the slowest open text-to-video model we've measured per second of footage. Pick it for its look, not its speed. The A100 80GB took almost exactly twice the H100's time (245 against 128 seconds), and the H200 was only 4% faster than the H100, so memory isn't the limit here; compute is.
About Mochi 1 Preview. Mochi 1 Preview: from genmo, 10B parameters, on Hugging Face since October 2024, Apache 2.0 licence. 9,119 downloads in the last 30 days and 1 community quantizations.
How it compares. H200: Mochi 1 Preview 0.4 frames/s, Wan 2.2 T2V A14B 0.35 (14B), Wan 2.1 T2V 14B 0.36 (14B), LTX-Video (distilled) 17.87 (2B), Wan 2.2 5B (720p) 1.41. 2 of 4 beat Mochi 1 Preview here.
Cost on a rented GPU. 1 minute of 24fps video of Mochi 1 Preview: $1.89 on a A100 80GB SXM4 ($0.95/hr, 2.0 hours), $3.61 on a H200 ($3.59/hr, 60 min, 1.9x the cost).
Mochi 1 Preview: cost per 1 minute of 24fps video on rented GPUs
| GPU | Cheapest rate | Speed (frames/s) | Cost per 1 minute of 24fps video |
|---|---|---|---|
| NVIDIA A100 80GB SXM4 | $0.95/hr | 0.2 | $1.89 |
| NVIDIA H100 80GB HBM3 | $2.14/hr | 0.38 | $2.24 |
| NVIDIA H200 | $3.59/hr | 0.4 | $3.61 |
Cheapest hourly rate we track on RunPod and Vast.ai, divided by the measured speed. Startup time and storage are extra.
Speed tiers for Mochi 1 Preview. under 1 frames/s: 3 (H200, H100 80GB HBM3, A100 80GB SXM4). 10 frames/s turns out a minute of 24fps video in under 2.5 minutes.
VRAM for Mochi 1 Preview. Measured peak 67.1GB, so 80GB is the smallest common card size; smallest card it ran on: H100 80GB HBM3 (80GB).
Power on Mochi 1 Preview. Most efficient: H200, 683W, 0.69 kWh per 1 minute of 24fps video. Hungriest: H100 80GB HBM3, 683W, 0.72 kWh. At $0.15/kWh: $0.10 per 1 minute of 24fps video.
Our verdict
Fastest on Mochi 1 Preview: NVIDIA H200, 0.4 frames/s. Cheapest to rent per job: NVIDIA A100 80GB SXM4, $1.89 per 1 minute of 24fps video.
FAQ
- What GPU do I need to run Mochi 1 Preview?
- About 70GB. Smallest card that ran it: NVIDIA H100 80GB HBM3 (80GB).
- How much does it cost to run Mochi 1 Preview in the cloud?
- $1.89 per 1 minute of 24fps video on a NVIDIA A100 80GB SXM4 at $0.95/hr, cheapest of 3 rentable cards we measured.
- Can I run Mochi 1 Preview on a 12GB, 16GB or 24GB card?
- It used 67.1GB at the precision we tested. 12GB: no; 16GB: no; 24GB: no.
- Is the H100 80GB HBM3 or the A100 80GB SXM4 faster for Mochi 1 Preview?
- The H100 80GB HBM3: 0.38 vs 0.2 frames/s, 91% faster on our bench.
How we test
Mochi 1 Preview in diffusers, bf16, one warmup clip and three timed clips from the same prompt, measured as frames of finished video per second, with CPU offload only on cards below the model's full-GPU size. Text-to-video is compute-bound and memory-hungry; cards that need CPU offload lose far more speed than their raw compute suggests.