Artificial Analysis vs SemiAnalysis

Anthropic Launches Claude Opus 5.5 Despite Slowdown Call

Anthropic has released Claude Opus 5.5, its flagship AI model and the first in the 5.5 family, achieving the highest score ever recorded on the Artificial Analysis Intelligence Index (58) and leading six out of ten benchmarks, including outperforming OpenAI's GPT-6 Astra on Terminal-Bench 4.0. Priced 20% lower for input/output tokens and 60% lower for cache reads, it claims a 40% cost reduction for typical workloads, though higher output token usage may offset savings; it's also 30% faster and now the default in Claude Code and the Claude app. OpenAI responded by releasing GPT-6 Sol and Luna, priced ~50% lower than predecessors and offering up to 90% discounts on cached input. Both cite efficiency gains. Claude Opus 5.5 is available on Claude apps, Claude Platform, AWS, Google Cloud, and Azure, with smaller models to follow.

• Claude Opus 5.5 scores 58 on the Artificial Analysis Intelligence Index, the highest recorded, and leads six out of ten benchmarks.
• Pricing is reduced: $4 per million input tokens, $20 per million output tokens, and cache reads at $0.20 per million tokens, promising a 40% cost reduction per task (though higher output token usage may offset savings).
• Opus 5.5 is 30% faster, supports multi-agent scaling up to 100 parallel agents, and is now default in Claude Code and the Claude app.

Xiaomi MiMo-V2.6-Pro tops open weights, trained for $3M

Xiaomi released MiMo-V2.6-Pro, a natively omnimodal open-weights model with 1.02T total / 42B active parameters, trained for $3M (about 130 hours and 75B tokens). It debuts as the top open-weights model on Artificial Analysis' Intelligence Index (46) with cost efficiency at $0.435/M input and $0.87/M output tokens, under an MIT license. Xiaomi also open-sourced the RL training environment code and recipes, but not the full 7k+ task datasets, signaling an emphasis on transparency in RL training.

• Xiaomi released MiMo-V2.6-Pro and MiMo-V2.6-Flash, natively omnimodal open-weights models.
• MiMo-V2.6-Pro has 1.02T total / 42B active parameters and tops the Artificial Analysis Intelligence Index at 46.
• RL training run cost $2.6M, used 75B tokens over 130 hours.

New Articles: Benchmarking GPT-6 Astra, Intelligence Index v4.3, and more

The articles page now shows 105 articles (up from 94), with new entries including 'Benchmarking GPT-6 Astra' (Sep 9, 2026), 'Announcing the Artificial Analysis Intelligence Index v4.3' (Sep 7, 2026), 'OpenBMB releases MiniCPM5-2B' (Sep 7, 2026), 'Announcing Artificial Analysis Intelligence Index v4.2' (Sep 4, 2026), 'Muse Spark 1.3: Meta reaches the frontier' (Sep 2, 2026), 'Google has released Gemini 3.8 Flash' (Sep 2, 2026), 'Claude Fable 5.1 tops the Artificial Analysis Intelligence Index' (Sep 1, 2026), 'Agnes AI releases Agnes 2.5 Pro Beta' (Aug 27, 2026), 'Intelligence at pocket scale' (Aug 24, 2026), 'Announcing the Speech Agent Arena' (Aug 24, 2026), and 'Announcing the Artificial Analysis Search Index' (Aug 18, 2026).

SemiAnalysis Tests Engram Offloading to DRAM and SSD

SemiAnalysis analyzes the Engram architecture, a model design that extends standard token embeddings with learned multi-token lookups, allowing for efficient parameter offloading to DRAM or SSD. This reduces HBM capacity requirements for models like DeepSeek-V4.1-Flash. Their experiments show that offloading Engram tables to DRAM can improve performance per dollar, while SSD offloading is currently not economically viable due to overhead. The article also benchmarks inference performance across NVIDIA and AMD GPUs, highlighting NVIDIA's CUDA moat and AMD's slower software support. The analysis includes findings on model behavior, such as gate scan results and ablation studies, and discusses the implications for HBM demand and model architecture innovation.

• SemiAnalysis tested offloading Engram tables to DRAM and SSD for DeepSeek-V4.1-Flash.
• Offloading to DRAM improved performance per dollar, reducing needed HBM capacity.
• SSD offloading was not economically viable, with DRAM delivering 121 million tokens per dollar versus 52 million for SSD.

SemiAnalysis Maps 300 US Datacenter Moratoriums, Finds Minimal Impact

SemiAnalysis published a detailed analysis of datacenter moratoriums in the US, arguing that the widespread narrative that these restrictions are killing the datacenter buildout is inaccurate. The analysis covers over 300 local moratoriums and four state-level actions (New York, Texas, Pennsylvania, Oregon). Using a project-by-project, parcel-level analysis of over 6,000 datacenters, the firm found that only approximately 1,525 MW of planned capacity is genuinely delayed by local moratoriums, representing 7.6% of the capacity sitting inside restricted boundaries. The firm's Datacenter Industry Model forecasts +38 GW of new US datacenter IT capacity in 2027, more than double 2026. The report also discusses public sentiment, finding that 46% of Americans view datacenters unfavorably, and examines the political dynamics driving moratoriums.

• SemiAnalysis maps over 300 local datacenter moratoriums and bans across the US, plus four state-level actions (New York, Texas, Pennsylvania, Oregon).
• Only 1,525 MW of planned capacity is directly delayed by local moratoriums, which is 7.6% of the ~20 GW exposed.
• SemiAnalysis forecasts +38 GW of US datacenter IT capacity delivered in 2027, more than double 2026.

Rubin NVL72 Agentic Inference: 67x Better Performance per Dollar

SemiAnalysis reports first verified agentic inference results for NVIDIA's Rubin NVL72 platform using their AgentX benchmark. Even on early pre-release software, Rubin delivers up to 67x better performance per dollar of TCO compared to GB300 in certain configurations, and significantly higher throughput per MW. The analysis projects Rubin can generate over 2x more profit per gigawatt than Blackwell, with revenue and profit advantages of 39% and 42% respectively at a fixed power budget. Dynamic power shifting (DSX MaxLPS) allows more GPUs per datacenter footprint. The article highlights Rubin's superiority over H200 and MI355X, with recommendations for inference providers to adopt Rubin for cost-efficient token generation.

• Rubin NVL72 achieves up to 67x the throughput per TCO of GB300 in specific scenarios.
• Rubin delivers up to 7x better token throughput per MW than Blackwell in real-world tests.
• Rubin can generate over 2x more profit per gigawatt than Blackwell.