DeepSeek reduces the AI performance gap with US rivals

First reported by Implicator.ai at · Updated · 2 sources

DeepSeek is within about 3% of the US leader on LiveBench, and its agentic coding score exceeds Anthropic's leading overall entry, according to Implicator.ai. DigiTimes reports that improvements in memory efficiency and closer work with Huawei suggest Chinese developers are finding ways to improve model performance despite tighter access to advanced US chips.

  • Implicator.ai says token traffic and profitability forecasts highlight commercial competition beyond benchmark scores.

Covered by 2 publishers within 25 hours of the first report.

Reporting2

Implicator.ai DeepSeek Narrows US AI Benchmark Lead to 3% After September Release · Marcus Schuler
DigiTimes DeepSeek closes in on US rivals despite chip curbs · DIGITIMES

Related

Topics AnthropicDeepSeekHuawei