China publishes 'landmark paper' on AI-to-AI technique that kicks human 'bottleneck' out of the loop and replaces us with an AI 'modem' — C2C brainwave direct connection achieves 150% boost in inference speed
Researchers from Tsinghua University have published a paper introducing a technique called Cache-to-Cache (C2C) that enables separate AI models to exchange internal information without generating text. This method bypasses the traditional bottleneck of text-based communication by directly transferring working memory between models, resulting in a 150% boost in inference speed. The paper, accepted at ICLR 2026, includes open-source code and utilizes a learned Fuser and selective gating to optimize data transfer.
- ▪Tsinghua University researchers developed the Cache-to-Cache (C2C) technique to allow AI models to share internal data without producing text.
- ▪The C2C method achieves a 150% improvement in inference speed by eliminating the time-consuming step of converting thoughts into written sentences.
- ▪The technique uses a learned Fuser to convert one model's internal data for another and employs selective gating to control which layers receive information.
- ▪The research paper describing C2C has been accepted at the ICLR 2026 conference and is accompanied by open-source code for developers.
TechRadar publishes from United Kingdom and files mainly under tech. We currently carry 2,822 of its stories.
Story provenance
Source · retrieval · rights · ranking — open for full record
inspect →
Story provenance
Attribution is not the same as permission. This drawer separates discovery metadata, excerpts, WeSearch-generated summaries, reuse status, and whether the publisher receives the visit. Nothing here claims a legal grant the publisher has not made.
Record
| Original publisher | TechRadar |
| Canonical URL | https://www.techradar.com/pro/china-publishes-landmark-paper-on-ai-to-ai-technique-that-kicks-human-bottleneck-out-of-the-loop-and-replaces-us-with-an-ai-modem-c2c-brainwave-direct-connection-achieves-150-boost-in-inference-speed |
| Publication time | Fri, 25 Sep 2026 00:15:00 +0000 |
| Retrieval time | 2026-09-25T00:20:26.870Z |
| Last seen | 2026-09-25T00:20:26.870Z |
| Headline source | Publisher (no WeSearch rewrite) |
| Excerpt source | publisher body |
| Excerpt method | First ~120 words (~800 chars) of extracted publisher body, fair-use limited. |
| Summary | WeSearch · cerebras-chat (WeSearch summarizer) |
| Summary source text | contentText |
| Citation coverage | Summary is a WeSearch-generated derivative; primary citation is the original publisher URL. |
| Cluster | eBGFdZsEdw4W · 1 stories |
| Cluster logic | Grouped by semantic title/content similarity across sources within a rolling window. Same-publisher template collisions are excluded from coverage comparison. |
| Ranking reason | Story pages are not engagement-ranked. Hub feeds use recency, with optional source-diversified chronological ordering (cap consecutive stories per source). No personalized ranking. |
| Publisher visit | Yes — open original |
| Substitutes article? | No — link-out required for full text |
Rights status (four layers)
WeSearch handling by dimension
| Indexing | May the item be indexed (stored, ranked, made findable)? | Allowed |
| Snippet | May a short excerpt of the publisher's text be shown? | Allowed |
| AI summary | May WeSearch generate its own short summary of the article? | Limited |
| Retrieval / RAG | May the content be exposed for third-party retrieval-augmented generation? | Not asserted |
| Model training | May the content be used to train AI models? | Not asserted |
| Commercial reuse | May the content be reused commercially? | Not permitted |
Basis: Derived from the published RSS/Atom feed. Contact: [email protected]. Reviewed: 2026-07-24.
Opening excerpt (first ~120 words) tap to expand
Pro China publishes 'landmark paper' on AI-to-AI technique that kicks human 'bottleneck' out of the loop and replaces us with an AI 'modem' — C2C brainwave direct connection achieves 150% boost in inference speed News By Efosa Udinmwen Published 25 September 2026 3 min read Text-based communication slows AI collaboration When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works. (Image credit: Yuichiro Chino/via Getty Images) Copy link Facebook X Whatsapp Reddit Pinterest Flipboard Threads Email Share this article 0 Join the conversation Follow us Add us as a preferred source on Google Newsletter Subscribe to our newsletter Cache-to-Cache lets separate AI models exchange internal information without generating textA learned Fuser converts one…
Excerpt limited to ~120 words for fair-use compliance. The full article is at TechRadar.
Discussion
0 commentsMore from TechRadar
