Dedicated GPU deployment
DeepSeek V4 Flash Vision Exp
Deploy DeepSeek V4 Flash Vision Exp on eight dedicated RTX PRO 6000 GPUs on one node, with text and image input and pre-downloaded weights.
Deploy with templateMeasured by EcoHash
RTX PRO 6000 benchmark
Results from the saved EcoHash run for this model configuration. The template may be updated independently; the methodology below states the tested settings.
- Time to first token
- 140 ms
- Single-request output speed
- 161.3 tokens/s
- Aggregate throughput
- 156.9 tokens/s
- End-to-end latency P50
- 2.98 s
- End-to-end latency P95
- 6.17 s
- Warm start
- 434.2 s
- Largest tested concurrency
- 1
- Throughput
- 17.4 outputs/min
- Success rate
- 100%
- Failures
- 0
- Out-of-memory failures
- 0
- GPU memory · loaded / idle
- 714.4 GB
- GPU memory · peak workload
- 714.4 GB
- Host RAM · peak
- 214.2 GB
- Weight storage
- 167.8 GB
- Cache footprint (downloads + runtime)
- 0 GB
- Output storage
- 0 GB
Test conditions
Main test: complete-answer C=1 run from the bastion through kubectl port-forward; public API and gateway excluded. TTFT p95 206.93 ms; TPOT p50/p95 6.20/7.22 ms/token. Output TPM 9,412.62 excludes input tokens. Cached-weight launch to Ready 434.16 s; cold-cache startup was not measured. Separate 500K multi-turn check: cold-prefix TTFT 94.392 s; continuation TTFT 1.705/1.558 s; 3/3 retrieval checks. Coding probe: 10 samples passed Python AST syntax checks; no formal quality score. Seed not fixed.
Full benchmark reportPublished template
Deployment configuration
Configuration from the published platform template. Review the template for current settings, availability, model terms and pricing.
- GPU
- 8 × NVIDIA RTX PRO 6000 Blackwell Server Edition
- Engine
- SGLang
- Deployment
- Dedicated model instance
- Billing
- GPU-hour · current price shown in the template
Guides and next steps
- Run DeepSeek with an EcoHash GPU template →
Review the quote, launch a template and connect to the resulting endpoint.
- Launch an 8-GPU RTX PRO 6000 instance →
Configure a whole-node workspace, SSH, an HTTP endpoint and persistent storage.
- Rent RTX Pro 6000 GPUs →
Compare 1/2/4/8-GPU configurations, live availability and per-second rental billing.
Deployment options
Opening a template does not start a paid instance. Review its configuration and price before launching.
Open the platform template to review its image, startup command, GPU configuration and current price before launching.