Unverified50% confidenceBenchmarkExact time
OpenAI's new model scored lower than the previous generation on two benchmarks: coding capability and computer operation.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/30/2026
First Seen
Valid until: 12/29/2026
Sources
Related Entities
Related Claims
Unverified在OpenAI的模型命名体系中，主版本号代表底层架构、训练范式或数据规模的代际升级，次版本号代表同一架构上的增量优化75% similarUnverifiedOpenAI offers Codex models with different parameter scales and capability tiers, where lightweight models are suitable for simple code completion and formatting tasks, and high-parameter models perform better in complex logical reasoning, multi-file refactoring, and architecture design.72% similarVerifiedThe OpenAI Codex desktop version is based on the codex-1 model, which has been fine-tuned with reinforcement learning and specifically optimized for code generation, debugging, and engineering tasks.72% similarUnverifiedOpenAI provides a one-click option in ChatGPT and Codex to downgrade to a lower-capability model for retry.72% similarUnverifiedAnthropic adopts a tiered model strategy of Haiku/Sonnet/Opus, while OpenAI has the o1-mini and o1-preview tiers.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/961369API
curl https://kongchang.com/api/v1/knowledge/claims/961369MCP
get_claim(id=961369)