AnswersPublished16 sources
AXP Explained: How a Deterministic Network Aims to Connect AI Agents
AXP is a platform designed to help AI agents find and connect with one another over more predictable network paths.
Published byEdited with GPT-6 LunaImages generated with GPT Image 2
Research answer
Key takeaways
- AXP is a platform designed to help AI agents find and connect with one another over more predictable network paths.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What is AXP, the AI agent interconnection platform unveiled by Purple Mountain Laboratories on October 16, 2026, at the 10th Future Network. Article summary: AXP is a deterministic-network-based platform intended to let AI agents from different vendors, organizations, and regions discover, connect to, and invoke one another. It was unveiled on October 16, 2026, at the 10th Fu. Topic tags: general, general web. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake numbers, click
AXP is a platform designed to help AI agents from different organizations and vendors discover and communicate with one another. Unveiled in Nanjing on October 16, 2026, it was jointly developed by Purple Mountain Laboratories, China Mobile, Jiangsu Future Network Group, and the Jiangsu Future Networks Innovation Institute. Its headline figures are developer-reported capabilities, not independently verified benchmarks.10
2
How AXP is intended to connect agents
AXP combines deterministic networking with semantic addressing, protocol translation, and coordinated computing and network-resource scheduling. The aim is to help agents find suitable counterparts and communicate across different systems and regions.10
Deterministic networking aims to make delivery more predictable within a controlled network domain. By contrast, ordinary best-effort internet connectivity does not promise fixed delivery times and can experience variable delays or packet loss. That distinction does not mean AXP can guarantee performance across every public-internet route; the cited descriptions refer to guarantees within a network domain.2
8
The launch materials describe the platform as able to translate between heterogeneous agent protocols and coordinate computing and network resources. They do not detail its discovery and invocation interfaces or list supported protocols and versions.10
2
AXP’s announced capacity and response targets
The developers announced that AXP can register up to one million agents and support 100,000 online and running simultaneously. That is a platform scale claim, not a published measure of how many tasks it can complete per second.10
Other reported capabilities include seconds-scale failure detection and switchover, and semantic-addressing responses on the order of 100 milliseconds. These figures describe separate functions; the addressing response time is not a claim that an entire agent task takes 100 milliseconds.8
10
What the CENI test results report
On CENI, China’s national future-network experimental facility, the team reported high-precision cross-domain agent addressing over more than 4,000 kilometers and a reduction of more than 20% in end-to-end task execution time.9
10
That reduction should be read as a reported result, not a fully documented independent benchmark. The cited launch coverage does not specify the comparison baseline or provide enough testing detail to reproduce the result.2
9 Some coverage uses different wording about scheduling latency, so it should not be conflated with the reported end-to-end task-time measure.
4
9
Testing, deployment reports, and open questions
Purple Mountain Laboratories says it has begun multi-scenario testing and industrial validation with telecom operators, communications-equipment companies, and large steel manufacturers.9
10 Separate reporting names Zhongtian Steel and describes applications at three locations in Jiangsu, as well as China Mobile deployments in 13 cities. Those reports do not establish that AXP is generally available for purchase.
5
Several details remain unclear in the cited coverage:
- “Industry first” status: The platform has been described as an industry first, but the available reports do not provide independent verification or a defined comparison with competing platforms.
2
8
- Benchmark methodology: The greater-than-20% task-time reduction is reported without a clearly specified baseline and testing methodology.
9
- Protocol compatibility: Heterogeneous protocol translation is an announced capability, but the launch reports do not confirm support for specific protocols such as MCP or A2A.
10
- Name and commercial terms: The cited materials do not explain what “AXP” stands for or establish a public release date, access process, or pricing.
AXP’s announcement describes an attempt to address both agent interoperability and the reliability of the network carrying agent interactions. Its reported scale and CENI results are notable claims, but protocol specifics, reproducible benchmarks, and public availability remain unresolved in the sources reviewed.
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "AXP Explained: How a Deterministic Network Aims to Connect AI Agents"?
AXP is a platform designed to help AI agents find and connect with one another over more predictable network paths.