Nvidia B200
PulseAugur coverage of Nvidia B200 — every cluster mentioning Nvidia B200 across labs, papers, and developer communities, ranked by signal.
- instance of graphics processing unit 90%
- instance of Beechcraft King Air 350 90%
- used by NVFP4 90%
- instance of GB200 90%
- competes with graphics card model series 80%
- competes with atom 80%
- competes with Readonflow Team 80%
- competes with NVIDIA H100 70%
- used by vLLM 70%
- competes with H200 70%
- used by CUDA 70%
- used by GB200 70%
20 day(s) with sentiment data
-
Revolut switches to self-hosted GLM on Nvidia B200s, cutting costs and improving speed
Revolut has replaced its use of Claude Code and Codex with a self-hosted General Language Model (GLM) running on 56 Nvidia B200 GPUs. This new setup is ten times cheaper, costing $0.30 per million tokens, and has improv…
-
Baidu's ERNIE Evolve system achieves new AI research benchmarks
Baidu has introduced ERNIE Evolve, an automated research system designed to optimize AI training and inference code. This system autonomously explores research directions, modifies code, and conducts experiments, retain…
-
AI market volatility: LLM prices fluctuate, GPU rentals rise, and funding news emerges · 1 source tracked
TANDIU's weekly data notes for November 2-9, 2026, highlight significant shifts in AI pricing and infrastructure. The LLM API list price for a specific model doubled and then halved within a week, underscoring the volat…
-
Backboard engine boosts GLM-5.3 output 5x over vLLM on B200 GPUs
Backboard's new inference engine reportedly achieved nearly five times the output of vLLM when using eight NVIDIA B200 GPUs with the GLM-5.3 model. This performance gain was observed under both sustained and burst load …
-
Backboard.io's new inference engine outperforms vLLM and SGLang
Backboard.io has developed a new inference engine that significantly outperforms existing solutions like vLLM and SGLang in terms of throughput and latency. Across various models including GLM-5.3, Qwen3.6-35B, and Qwen…
-
New LEGIT system enhances constrained decoding for generative recommenders
Researchers have developed a new method called LEGIT (Live Eligibility Gating over Identifier Tries) to improve constrained decoding for generative recommenders, particularly for applications like the Uber Eats search e…
-
LLM context length limited by hardware memory, not model size
A recent analysis reveals that the ability of large language models to handle long contexts is primarily determined by the hardware's memory capacity rather than the model's architecture or quantization methods. The cri…
-
Self-hosting LLMs to cost $150K-$2M+ annually in 2026
Self-hosting open-source large language models in 2026 is projected to cost between $150,000 and over $2 million annually. The significant expenses extend beyond GPU hardware to include the cost of specialized engineeri…
-
LLM VRAM Needs: Llama, Qwen, and DeepSeek Self-Hosting Guide
Self-hosting large language models requires careful consideration of VRAM needs, which are influenced by model weights and KV cache size. Dense models like Llama scale predictably, with the 70B version demanding signifi…
-
Cost analysis reveals fine-tuning 70B models ranges from $1-2 per million tokens
Fine-tuning a 70 billion parameter model can cost between $1 and $2 per million training tokens, with a 20 million token LoRA run estimated at around $20. The cost is primarily determined by the number of tokens trained…
-
AutoCompress framework automatically shrinks LLMs for specific hardware
A new framework called AutoCompress has been developed to automatically compress large language models (LLMs) while adhering to specific hardware and performance constraints. This hardware-guided approach uses different…
-
Looped transformers' HBM traffic savings on Nvidia B200 analyzed
A recent analysis explores the efficiency of looped transformer architectures, specifically the Huginn model, in managing High Bandwidth Memory (HBM) traffic on Nvidia's B200 hardware. The study found that while sharing…
-
HyperFlow H3 Flash model generates video and audio with fewer steps
Video Rebirth has released HyperFlow H3 Flash, a compressed version of MiniMax-H3 that generates synchronized video and stereo audio. This model, available in 14B parameter sizes, can produce 5.2 to 15-second videos at …
-
Revolut slashes AI coding costs by 90% with self-hosted GLM on Nvidia B200s
Revolut has significantly reduced its AI coding costs by migrating approximately 2,000 developers from paid APIs to a self-hosted GLM model. This transition, utilizing 56 Nvidia B200 GPUs, has decreased token costs by a…
-
Together AI unveils Polymorphic Inference for adaptable AI model deployment
Together AI has introduced a new research agenda called Polymorphic Inference, designed to address the rapid evolution of AI models, architectures, and hardware. This approach contrasts with traditional Static Inference…
-
AI chip smugglers exploit loophole to bring banned HBM into China
Smugglers are exploiting a loophole in trade controls to bring banned High Bandwidth Memory (HBM) chips from South Korea into China. These expensive HBM chips, crucial for AI GPUs, are paired with cheaper, less regulate…
-
Mistral Large 4 open weights released; self-hosting requires substantial hardware
Mistral AI has released the open weights for its Mistral Large 4 model, a 1.05 trillion parameter mixture-of-experts architecture. This model, nicknamed "Le Chonk," features a 1 million token context window and supports…
-
NVIDIA H200 GPU pricing offers 15% cost increase for 43% more bandwidth
A comparison of GPU pricing for AI workloads reveals that while NVIDIA's H200 offers a significant bandwidth increase over the H100, its cost is only moderately higher. Specifically, the H200 is priced 15% more than the…
-
NVIDIA H200 GPU offers cost advantages over H100 for memory-bound AI tasks
The NVIDIA H200 GPU offers improved memory bandwidth compared to the H100, making it more efficient for memory-bound tasks like LLM inference. While the H200's compute power is similar to the H100, its higher memory thr…
-
AI agent autonomously optimizes NVIDIA GPU kernel, beats baseline by 9.8%
An AI agent named ForgeKernel has autonomously optimized NVIDIA's FlashMLA baseline for the B200 GPU, achieving a 9.8% performance increase in under 22 hours. The agent was able to improve the kernel's speed from 518 µs…