Multiverse Computing Unveils Breakthrough: All CompactifAI Models Now Run on Intel Xeon 6 Processors

Advancement delivers significant performance improvements, energy savings, and reductions in memory footprint while preserving accuracy

SAN SEBASTIÁN, Spain, July 23, 2026 (GLOBE NEWSWIRE) -- Multiverse Computing, a leader in AI model compression, today announced that its CompactifAI^®-compressed version of the Llama* 3.3 70B model now runs seamlessly on Intel^® Xeon^® 6 processors with Performance-cores, utilizing vLLM* CPU and Intel^® Advanced Matrix Extensions (AMX), marking a major advancement in more energy-efficient and scalable AI . With CompactifAI, Multiverse Computing’s proprietary AI model compression technology, the company compresses and offers optimized versions of leading open-source AI models. This breakthrough delivers improved throughput and latency while maintaining compatibility with mainstream open‑source tooling.

Performance and efficiency. In controlled benchmarks using an Intel^® Xeon^® 6737P processor, the CompactifAI-compressed Llama* 3.3 70B model demonstrated significant improvements over the uncompressed baseline. For a request comprising 1,024 input and 1,024 output tokens, the uncompressed baseline achieved an output throughput of 2.00 tokens/second and token throughput of 4.02 tokens/second. The CompactifAI-compressed model delivered an output throughput of 3.86 tokens/second and a total token throughput of 7.81 tokens/second, marking improvements of 93.6% and 94.1%, respectively.

Latency improvements were observed across all concurrency levels. For example, at one concurrent user, the uncompressed baseline required 5,056.34 seconds to process, while the compressed model reduced this to 2,598.22 seconds, resulting in a latency reduction of 48.6%. This trend continued across higher concurrency levels, with the compressed model consistently delivering faster response times.

Additional performance metrics including ITL (Inter-Token Latency), TPOT (Time Per Output Token), and TTFT (Time to First Token) showed substantial reductions:

• ITL (mean): 493.36 ms to 252.08 ms (48.9% reduction).
• TPOT (mean): 494.71 ms to 255.76 ms (48.3% reduction).
• TTFT (mean): 6939.20 ms to 3705.64 ms (46.6% reduction).
  

Across all concurrency levels tested, the compressed model showcased consistent latency reductions and throughput improvements, with the largest gains at the highest concurrency level (256 concurrent users), where throughput increased by 107.0% and latency decreased by 51.7%.

Accuracy. When measured against standard benchmarks, the CompactifAI-compressed model retained strong accuracy relative to the baseline model, with only minor variations observed:

• BoolQ*: Baseline score of 0.904587 vs. compressed score of 0.896024 (0.95% decrease).
• GSM8K*: Baseline score of 0.934799 vs. compressed score of 0.924185 (1.14% decrease).
• HellaSwag*: Baseline score of 0.590520 vs. compressed score of 0.579068 (1.94% decrease).
• MMLU*: Baseline score of 0.776528 vs. compressed model score of 0.757300 (2.48% decrease).
• WinoGrande* (Template): Baseline score of 0.678769 vs. compressed score of 0.725335 (6.86% increase).
  

While it may seem counterintuitive, the accuracy improvement observed in some benchmarks can be attributed to the re-training, or “healing” process that follows compression. During compression, Multiverse Computing reduces the model’s parameter count while simultaneously applying a targeted re-training phase to optimize performance. Although improvement across all benchmarks cannot be guaranteed, the compressed model retains a significant share of the original model’s core information, which can in select cases yield performance gains, as demonstrated by the 6.86% improvement observed in the WinoGrande benchmark..

Overall, the compressed model retained over 97% of the baseline accuracy across these benchmarks, confirming that the compression did not meaningfully affect performance for real-world applications.        

Model size on disk. The compressed model's disk size is reduced by approximately 50%, from ~130 GiB to ~65 GiB, significantly reducing storage requirements and provisioning times for large-scale deployments.

According to Enrique Lizaso, Co‑founder & CEO of Multiverse Computing, "This isn’t just a technical milestone — it’s a game‑changer for AI builders."

Built for real‑world deployment. CompactifAI integrates with PyTorch* software, Hugging Face* platform, and other open‑source frameworks, providing a straightforward path into existing developer pipelines. On the infrastructure side, the stack supports enterprise‑grade AI workloads such as retrieval‑augmented generation (RAG), multimodal reasoning, and domain‑specific applications. Supported models include Llama* 4 Scout, Llama* 3.3 70B, Llama* 3.1 8B, Mistral* Small 3.1, and DeepSeek* R1, along with ‘Slim’ variants—compressed versions of the original AI models optimized by CompactifAI for enhanced efficiency.         

Availability. CompactifAI is available across leading cloud platforms, and now efficiently supports all instances powered by Intel^® Xeon^® 6 processors with Performance-cores running vLLM* CPU, complementing on-premises deployments and providing a scalable, secure path to production. The solution is optimized for demanding enterprise applications in industries such as finance, healthcare, and manufacturing, offering both high performance and energy efficiency.

Editor’s note: A brief performance highlight appears in the main text. For a full datacentre perspective, multi‑user concurrency testing, complete configuration details, and measurement methodology, see Appendix A and Figure 1 below.

Figure 1 — Server-side throughput (tokens/second) and latency (ms) measured for both the uncompressed baseline and CompactifAI-compressed variants. Results were obtained using a test configuration of one concurrent user, 1,024 input tokens, and 1,024 output tokens, performed on an Intel^® Xeon^® 6 processor with Performance-cores usingvLLM* CPU at tensor parallelism level 1. The figures compare baseline (uncompressed) and CompactifAI (compressed) model performances across throughput (tokens per second) and latency (milliseconds). See Appendix A for methodology.

To learn more, visit multiversecomputing.com/compactifai. For inquiries, contact us at sales@multiversecomputing.com.

Appendix A — Reproducibility & Methodology

Hardware & Firmware

• CPUs: 1× Intel Xeon 6737P (32 physical cores, 64 hardware threads, up to ~4.0 GHz). Link to specifications: Intel® Xeon® 6737P Processor.
• System memory: ~1.0 TB RAM.
  
  

Software Stack

• Serving: vLLM* (CPU build) running inside a Docker* container (vLLM*-based serving image).
• Models:
  • Compressed: CompactifAI variants served from local checkpoints.
  • Baseline: Corresponding baselines, served from standard HF checkpoints.
  • Python* environment: Dedicated environment for the vLLM* client and analysis scripts (benchmark driver, metric aggregation, plotting).
    
    

Workload Definition

• Prompt/completion lengths: 1,024 input tokens and 1,024 output tokens, using a synthetic random-token workload.
• Concurrency levels: 1 (low-load latency) - 256 (high-concurrency stress test, representative of datacentre-style load).
• Request volume: For each configuration, approximately 10×concurrency requests are issued (e.g., ~2,560 requests at concurrency 256).
• Reported metrics (per model/config):
  • Overall benchmark duration (s).
  • Request throughput (req/s), total-token and output-token throughput (tokens/s).
  • Latency metrics: mean TTFT (Time To First Token), mean TPOT (Time Per Output Token, excluding the first), and mean ITL (Inter-Token Latency), with additional distribution details available in the raw JSON outputs.
    
    

About Intel

Intel is an industry leader, creating world-changing technology that enables global progress and enriches lives. Inspired by Moore’s Law, we continuously work to advance the design and manufacturing of semiconductors to help address our customers’ greatest challenges. By embedding intelligence in the cloud, network, edge and every kind of computing device, we unleash the potential of data to transform business and society for the better.

About Multiverse Computing

Multiverse Computing is the leader in AI model compression. The company’s deep expertise in quantum software led to the development of CompactifAI, a revolutionary compressor that reduces computing requirements and unleashes new use cases for AI across industries. Headquartered in Donostia, Spain, with offices in the United States, Canada, and across Europe, Multiverse serves more than 100 global customers, including Iberdrola, Bosch, and the Bank of Canada. For more information, visit www.multiversecomputing.com.

Media Contact

LaunchSquad for Multiverse Computing

multiverse@launchsquad.com

Disclaimer:

For Intel products, performance varies by use, configuration and other factors. Learn more on the Performance Index site. Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available updates. See backup for configuration details.

Intel technologies may require enabled hardware, software or service activation. Your costs and results may vary. No product or component can be absolutely secure. Intel does not control or audit third-party data. You should consult other sources to evaluate accuracy.

*Other names and brands may be claimed as the property of others.
© Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries.

Photos accompanying this announcement are available at: 
https://www.globenewswire.com/NewsRoom/AttachmentNg/c954b3d1-f586-44d6-a539-9caf8865a773
https://www.globenewswire.com/NewsRoom/AttachmentNg/2216359e-3d14-41e4-ba92-3ef07e830f35
https://www.globenewswire.com/NewsRoom/AttachmentNg/8dc7cd2a-c978-416f-a058-06a6e0be00d2
https://www.globenewswire.com/NewsRoom/AttachmentNg/b8a1382b-0981-41ec-af55-a96b71c51824
https://www.globenewswire.com/NewsRoom/AttachmentNg/5d640fd9-b71e-43e9-be8a-ec3ddf8ccd82