GPT‑5.4
StudentSim outperforms GPT‑5.4 in chess, second‑language English writing, and mathematics Mistral Large 4 achieves a score on DeepSWE v1.1 MetaMask introduced Agent Wallet in August 2026 MetaMask’s Agent Wallet includes Guard Mode as a feature GLM-5 (Baseten) matches GPT-5.4 on core agent tasks. MiniMax M2.7 (OpenRouter) matches GPT-5.4 on core agent tasks. GLM-5 (Baseten) has a latency of 0.65 seconds, compared to 2.56 seconds for GPT-5.4. GPT-5.4 has a latency of 2.56 seconds, compared to 0.65 seconds for MiniMax M2.7 (OpenRouter). GLM-5 (Baseten) has a throughput of 70 tokens per second, compared to 34 tokens per second for GPT-5.4. GPT-5.4 has a throughput of 34 tokens per second, compared to 70 tokens per second for MiniMax M2.7 (OpenRouter). GLM-5 (Baseten) charges $0.95 per million tokens for input, compared to $2.50 for GPT-5.4. MiniMax M2.7 (OpenRouter) charges $0.30 per million tokens for input, compared to $2.50 for GPT-5.4. GLM-5 (Baseten) charges $3.15 per million tokens for output, compared to $15.00 for GPT-5.4. MiniMax M2.7 (OpenRouter) charges $1.20 per million tokens for output, compared to $15.00 for GPT-5.4. COBOL migration benchmark translates COBOL to Python GPT-6.1 Sol translates COBOL to Python GPT-5.4 is used in Novelty Bench evaluation· Graphiti 生成的实体摘要
关系变化
按生效时间- GPT‑5.4USED INNovelty Bench
GPT-5.4 is used in Novelty Bench evaluation
- GPT-6.1 SolTRANSLATES COBOL TOGPT‑5.4
GPT-6.1 Sol translates COBOL to Python
- COBOL migrationTRANSLATES COBOL TOGPT‑5.4
COBOL migration benchmark translates COBOL to Python
- GLM-5 (Baseten)HAS OUTPUT PRICEGPT‑5.4
GLM-5 (Baseten) charges $3.15 per million tokens for output, compared to $15.00 for GPT-5.4.
- GLM-5 (Baseten)HAS THROUGHPUTGPT‑5.4
GLM-5 (Baseten) has a throughput of 70 tokens per second, compared to 34 tokens per second for GPT-5.4.
- MiniMax M2.7 (OpenRouter)HAS OUTPUT PRICEGPT‑5.4
MiniMax M2.7 (OpenRouter) charges $1.20 per million tokens for output, compared to $15.00 for GPT-5.4.
- GLM-5 (Baseten)HAS LATENCYGPT‑5.4
GLM-5 (Baseten) has a latency of 0.65 seconds, compared to 2.56 seconds for GPT-5.4.
- GPT‑5.4HAS LATENCYMiniMax M2.7 (OpenRouter)
GPT-5.4 has a latency of 2.56 seconds, compared to 0.65 seconds for MiniMax M2.7 (OpenRouter).
- GLM-5 (Baseten)MATCHESGPT‑5.4
GLM-5 (Baseten) matches GPT-5.4 on core agent tasks.
- MiniMax M2.7 (OpenRouter)HAS INPUT PRICEGPT‑5.4
MiniMax M2.7 (OpenRouter) charges $0.30 per million tokens for input, compared to $2.50 for GPT-5.4.
- GLM-5 (Baseten)HAS INPUT PRICEGPT‑5.4
GLM-5 (Baseten) charges $0.95 per million tokens for input, compared to $2.50 for GPT-5.4.
- GPT‑5.4HAS THROUGHPUTMiniMax M2.7 (OpenRouter)
GPT-5.4 has a throughput of 34 tokens per second, compared to 70 tokens per second for MiniMax M2.7 (OpenRouter).
- MiniMax M2.7 (OpenRouter)MATCHESGPT‑5.4
MiniMax M2.7 (OpenRouter) matches GPT-5.4 on core agent tasks.
- StudentSimOUTPERFORMSGPT‑5.4
StudentSim outperforms GPT‑5.4 in chess, second‑language English writing, and mathematics
- Mistral Large 4ACHIEVES SCORE ONGPT‑5.4
Mistral Large 4 achieves a score on DeepSWE v1.1
- GPT‑5.4HAS FEATUREGPT-Live-1.1
MetaMask’s Agent Wallet includes Guard Mode as a feature
- GPT‑5.4INTRODUCEDGPT-OSS-20B
MetaMask introduced Agent Wallet in August 2026
关系是 Graphiti 从报道中提取的事实陈述，不是原文引述；失效时间表示后来的报道显示这条关系已不成立。
相关事件 4
来源报道 6
- 我们能信任 LLM 判断研究创新性吗？Hugging Face Blog · 一手 · 2026-10-28
- OpenAI 称 GPT-6 遗留语言编程能力显著提升The Decoder · 2026-10-21
- OpenAI’s rogue crypto agent spent funds without permission, exposing risks to autonomous walletsCryptoSlate · 2026-10-17
- Open Models Have Crossed a ThresholdLangChain Blog · 2026-10-14
- 模拟学生学习，提升AI辅导效果Microsoft Research · 一手 · 2026-10-13
- Mistral AI 发布 Mistral Large 4 预览版Mistral AI · 一手 · 2026-10-06