Unverified50% confidenceBenchmarkExact time
Hydrogen leads on the OS World 2.1 offline computer-use test, and leads Sonnet and GPT 6.1 Tera on Agent's Last Exam and Multimodal Knowledge Agent Benchmark, respectively.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
11/15/2026
First Seen
Valid until: 11/29/2026
Sources
Related Entities
Related Claims
UnverifiedIn the official ten tests, Hydrogen took first place in seven against lightweight models such as Cloud Sonnet 5.5 and GPT Terra 6.1, with a comprehensive index score of 51 at the default level, surpassing its own Argon model.69% similarUnverifiedGoogle released the mid-tier model Hydrogen, benchmarked against Claude Sonnet and ChatGPT Terra, characterized by fast execution speed.66% similarUnverifiedGrok still has a certain gap compared to GPT-4o and Claude 3.5 Sonnet on programming-specific benchmark tests such as HumanEval and MBPP.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1172020API
curl https://kongchang.com/api/v1/knowledge/claims/1172020MCP
get_claim(id=1172020)