Unverified50% confidenceFactExact time
Axion 1.0's openness extends beyond the model weights, with the complete recipe released as well, including data construction, generation, filtering, validation, prompts, and training code.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/4/2026
First Seen
Valid until: 2/2/2027
Sources
Related Entities
Related Claims
UnverifiedOfficials say Axion 1.0 achieved strong performance on a broad set of agent benchmarks and is competitive with leading commercial models on several of them.69% similarVerifiedFrontierCode 1.1's test cases are typically derived from real open-source projects and require models to have cross-file context understanding and long-sequence reasoning capabilities.66% similarUnverifiedThe signals produced by this method can flag problematic training samples that frontier model classifiers would miss.65% similarUnverifiedCompared with runnable prototypes, production-grade code additionally requires thorough error handling and edge cases, unit and integration test coverage, security, performance and scalability, observability, and documentation conforming to team standards.64% similarUnverifiedThe research team made all code, main result weights, and evaluation scripts publicly available; deep variants provide only code without trained checkpoints.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1135500API
curl https://kongchang.com/api/v1/knowledge/claims/1135500MCP
get_claim(id=1135500)