Jul 4
This looks really awesome. Many flops, much #NVIDIA B300s, deep infrastructure.
5
15
95
30,122
Fast ML inference. Run top AI models using a simple API.
Joined February 2023
- Posts 675
- Following 65
- Followers 5,299
- Likes 202
91 Photos and videos
Jul 4
We are excited to be part of the largest infrastructure build out. I feel in some sense this is similar to the undersea cables carrying the internet traffic. Similar to the interstate highway system, similar to the railway networks. We enjoy building things!
2
15
1,509
Jul 2
Text to video in ~2.4 seconds.
FastWan-QAD-FP8-1.3B is now on DeepInfra 🆕
5-second 480p clips from a prompt, landscape or portrait. 3-step inference, $0.0025/second.
3
3
17
1,108
Jul 2
🆕🚀Batch API is live on DeepInfra.
Run large inference jobs → embeddings, evals, classification, summarization — asynchronously at 20% off real-time pricing.
OpenAI-compatible. If you've used OpenAI's Batch API, your existing code just works. Change base_url, done.
Results back within 24h 👇
1
4
10
865
Jul 1
New on DeepInfra: Priority Service Tier.⚡️
Your latency-critical traffic jumps to the front of the queue, lower TTFT, protected admission under load. One field: service_tier="priority" 1.5× the real-time price. You only pay if priority is actually delivered.
2
3
10
898
Jun 30
We're proud to serve our models on @nvidia inference stack 🚀
NVIDIA inference software keeps driving down token costs, long after AI infrastructure is deployed. ⚡
In just one month on NVIDIA Blackwell, software optimizations improved DeepSeek V4 performance by up to 5×, reducing token costs to roughly one-fifth of previous levels. NVIDIA's integrated inference software stack compounds improvements across runtimes, kernels, networking, and hardware, delivering up to 20× higher throughput on the same GPU.
Co-designed with NVIDIA GPUs, CPUs, networking, and systems, and powered by CUDA-native open source frameworks, NVIDIA's inference software stack ensures new model breakthroughs and optimizations run on NVIDIA from day zero, and keep improving throughput and lowering cost after deployment.
See how @Baseten, @Cognition, @DeepInfra, @togethercompute, and @Cursor_ai are turning continuous software innovation into lower cost per token: nvda.ws/4eRT43m
1
1
11
1,877
Jun 30
We’ve been building on @nvidia's inference stack from day one — TensorRT-LLM, Dynamo, NVFP4 on Blackwell.
When DeepSeek V4 dropped, we served it on day 0. Then we moved to B300 after measuring a 4x perf increase. A workload that needed 4×H200 now runs on a single B300.
blogs.nvidia.com/blog/infere…
1
5
618
Jun 25
We keep thinking "How did she do it?"
🎬 Just shipped: LTX-2.3-Distilled-Diffusers is live on Deep Infra.
1080p. 5 seconds of video. ~24 seconds on @nvidia Blackwell.
The model is live now at $0.035/second.
Stay tuned for speed improvements.
2
1
11
536
Jun 25
Pay attention to the prompt details: "A woman in her twenties dressed as a friendly party clown — light, natural face makeup, a round red nose, bright red curly clown wig, and a colorful costume — performs a magic trick on the green lawn of a cozy backyard birthday party: with a graceful flourish of her hands, a bright bouquet of colorful flowers suddenly appears, and a small group of excited kids gasp, clap, and laugh. Natural human face with realistic skin texture and a warm genuine smile. Behind them: a wooden fence, a picnic table with a frosted birthday cake, and colorful decorations strung between two trees. Bright natural afternoon sunlight, candid documentary style, photorealistic, shot on a DSLR, vivid natural colors, lively gentle motion, sharp focus."
Try it yourself → deepinfra.com/FastVideo/LTX-…
1
2
290
Jun 22
@Zai_org GLM 5.2 = GPT-5.5 coding performance
at 1/6th the cost.
open weights.
now $1 / $4 / $0.18 per 1M tokens
best price on the market 🔥
1
3
14
838
Jun 22
1
1
255
Jun 18
GLM 5.2 prices just dropped 🔥
$1.40 → $1.20 input
$4.40 → $4.20 output
$0.25 → $0.20 cached
cached tokens down 20% - huge for long-context workloads.
same model, less spend. 👇
2
1
10
949
Jun 18
1
1
540
Jun 18
Your own AI agent, always on, from $13/mo. 📷
Deep Infra Hosted Agents: OpenClaw (web dashboard) or Hermes (SSH/terminal). One-click setup & updates. Pre-wired for fast inference from second one. Auto-backups restore. Stop it and pay $0 while idle.
Spin one up → deepinfra.com/dash/agents
2
2
15
1,227
Jun 16
📉 Price cut on @NVIDIAAI Nemotron 3 Ultra.
$0.50 in / $2.20 out / $0.10 cached per 1M — output down 12%, cached down 33%.
550B/55B MoE for agentic reasoning and deep research. Multimodal, function calling, 256K context.
deepinfra.com/nvidia/NVIDIA-…
2
1
7
414