Dolphin X1 Trinity Nano 6B · 11 GPUs measured first-party · llama.cpp · Updated October 2026
What GPU Do You Need for Dolphin X1 Trinity Nano 6B?
Dolphin X1 Trinity Nano 6B on 11 GPUs, measured first-party: NVIDIA RTX PRO 6000 Blackwell Workstation Edition leads at 282.0 tok/s, T4 trails at 114 tok/s, and it peaked at 5GB of VRAM.
Benchmarked weights: mradermacher/Dolphin-X1-Trinity-Nano-GGUF
NVIDIA RTX PRO 6000 Blackwell Workstation Edition
282.0 tok/s on Dolphin X1 Trinity Nano 6B, the ceiling. Measured on our bench. 96GB of VRAM, $8,565 at launch.
- 282.0 tok/s on Dolphin X1 Trinity Nano 6B
- 96GB, clears the Dolphin X1 Trinity Nano 6B floor
- Rentable by the hour rather than bought
- 600W board rating
- Datacenter or workstation hardware, not a retail purchase
What GPU Do You Need for Dolphin X1 Trinity Nano 6B?, tok/s by GPU
Efficiency: tok/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: tok/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.
Dolphin X1 Trinity Nano 6B. Measured tokens per second by GPU
| GPU | tok/s | Prompt t/s | tok/W | Avg power |
|---|---|---|---|---|
| NVIDIA RTX PRO 6000 Blackwell Workstation Edition | 282 | 15144.4 | 2.28 | 123.6 W |
| NVIDIA H200 | 259.8 | 13347.2 | 1.8 | 144.3 W |
| NVIDIA B300 | 257.9 | 10354 | 1.04 | 248.6 W |
| NVIDIA H100 80GB HBM3 | 247.9 | 12952.1 | 1.76 | 140.8 W |
| NVIDIA L40S | 246.4 | 16432.9 | 2.09 | 117.8 W |
| NVIDIA B200 | 209.3 | 13149.7 | 0.77 | 271.5 W |
| NVIDIA A10G | 171.2 | 6834 | 1.84 | 93.1 W |
| NVIDIA L4 | 166.6 | 7451.3 | 3.53 | 47.2 W |
| NVIDIA A100 80GB SXM4 | 158.4 | 7111.9 | 1.21 | 130.7 W |
| NVIDIA A100 40GB SXM4 | 153.3 | 6841.7 | 1.69 | 90.8 W |
| NVIDIA T4 | 114.2 | 3118.4 | 2.13 | 53.5 W |
What the numbers show. Across 11 GPUs measured on our own bench, RTX PRO 6000 Blackwell Workstation Edition is fastest at 282 tok/s. The slowest, T4, manages 114, so the spread is 2.5x from top to bottom. L4 is the most efficient, 167 tok/s at 47W. Per dollar of launch price, L4 gives the most (66.6 tok/s per $1,000). The fastest card with 16GB or less is T4 at 114 tok/s.
How it compares. H100 80GB HBM3: Dolphin X1 Trinity Nano 6B 247.9 tok/s, DeepSeek-R1-0528-Qwen3-8B 245.3 (8B), Josiefied-Qwen3-8B-abliterated-v1 245.0, Qwen3 8B 244.2 (8B), Qwen3-8B 244.2. Dolphin X1 Trinity Nano 6B beats all 4 here.
Cost on a rented GPU. 1M generated tokens of Dolphin X1 Trinity Nano 6B: $0.33 on a T4 ($0.14/hr, 2.4 hours), $1.06 on a RTX PRO 6000 Blackwell Workstation Edition ($1.08/hr, 59 min, 3.2x the cost).
Dolphin X1 Trinity Nano 6B: cost per 1M generated tokens on rented GPUs
| GPU | Cheapest rate | Speed (tok/s) | Cost per 1M generated tokens |
|---|---|---|---|
| NVIDIA T4 | $0.14/hr | 114.2 | $0.33 |
| NVIDIA L4 | $0.44/hr | 166.6 | $0.73 |
| NVIDIA A100 40GB SXM4 | $0.47/hr | 153.3 | $0.86 |
| NVIDIA L40S | $0.79/hr | 246.4 | $0.89 |
| NVIDIA RTX PRO 6000 Blackwell Workstation Edition | $1.08/hr | 282 | $1.06 |
| NVIDIA A100 80GB SXM4 | $0.95/hr | 158.4 | $1.66 |
| NVIDIA H100 80GB HBM3 | $2.14/hr | 247.9 | $2.39 |
| NVIDIA H200 | $3.59/hr | 259.8 | $3.84 |
| NVIDIA B300 | $6.94/hr | 257.9 | $7.47 |
| NVIDIA B200 | $5.98/hr | 209.3 | $7.94 |
Cheapest hourly rate we track on RunPod and Vast.ai, divided by the measured speed. Startup time and storage are extra.
Speed tiers for Dolphin X1 Trinity Nano 6B. 30+ tok/s: 11 (RTX PRO 6000 Blackwell Workstation Edition, H200, B300). 30 tok/s is roughly where replies outpace reading.
Reading your prompt. Before Dolphin X1 Trinity Nano 6B writes anything it reads the input: 16432.9 tok/s on the L40S (0.2s for a 4,000-token prompt), 3118.4 on the T4 (1.3s). Long documents and big code files feel this number more than the generation speed.
VRAM for Dolphin X1 Trinity Nano 6B. Measured peak 4.4GB, so 8GB is the smallest common card size; smallest card it ran on: T4 (16GB). With long context: Q4_K_M 5GB (tested), Q2_K 4GB, Q3_K_M 4GB, Q5_K_M 6GB, Q6_K 7GB.
Power on Dolphin X1 Trinity Nano 6B. Most efficient: L4, 47W, 78.7 Wh per 1M generated tokens. Hungriest: B200, 272W, 0.36 kWh. At $0.15/kWh: $0.012 per 1M generated tokens.
Our verdict
Fastest on Dolphin X1 Trinity Nano 6B: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 282.0 tok/s. Cheapest to rent per job: NVIDIA T4, $0.33 per 1M generated tokens.
FAQ
- What GPU do I need to run Dolphin X1 Trinity Nano 6B?
- About 5GB. Smallest card that ran it: NVIDIA T4 (16GB).
- How much does it cost to run Dolphin X1 Trinity Nano 6B in the cloud?
- $0.33 per 1M generated tokens on a NVIDIA T4 at $0.14/hr, cheapest of 10 rentable cards we measured.
- Can I run Dolphin X1 Trinity Nano 6B on a 12GB, 16GB or 24GB card?
- It used 4.4GB at the precision we tested. 12GB: yes; 16GB: yes; 24GB: yes.
- Is the H100 80GB HBM3 or the A100 80GB SXM4 faster for Dolphin X1 Trinity Nano 6B?
- The H100 80GB HBM3: 247.9 vs 158.4 tok/s, 57% faster on our bench.
How we test
llama.cpp llama-bench at Q4_K_M, 512-token prompt and 128 generated tokens, three runs after a warmup, full GPU offload, with power and VRAM sampled throughout. Token generation is memory-bandwidth-bound, so the ranking tracks bandwidth closely, which makes it a fair guide to cards we haven't run yet.