Expired50% confidenceFactExact time
Grok 4.5 outputs at 80 tokens per second and features a context window of up to 500,000 tokens (approximately 400,000 English words).
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/24/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 10/22/2026(expired)
Sources
Grok 4.5深度解析：专为编程与Agent而生的工程大脑
bilibili傻狗与AI7/9/2026
Related Claims
UnverifiedGrok 4.5 is in the fast model tier and can achieve an output speed of approximately 80 TPS (tokens per second).79% similarUnverifiedGroq's LPU focuses on ultra-low latency inference and has achieved output speeds exceeding 500 tokens per second.74% similarUnverifiedGrok 4.5 currently supports a 500K token context window and is reportedly set to be upgraded to 1 million tokens in the near future.74% similarUnverifiedIn prefill speed tests, Halo achieved approximately 650 tokens/s while Spark achieved approximately 2000 tokens/s, with Spark being nearly three times faster.71% similarUnverifiedGrok 4.5支持128K Token的上下文长度70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/607240API
curl https://kongchang.com/api/v1/knowledge/claims/607240MCP
get_claim(id=607240)