Verified70% confidenceFactExact time
Even when the temperature is set to 0, some inference frameworks still have non-determinism in underlying floating-point computations, causing output differences across hardware and across batches.
4
Sources
70%
Confidence
Long-term
Relevance
7/16/2026
First Seen
Sources
Related Claims
VerifiedThe non-determinism of large language models stems from their autoregressive sampling process; even with temperature set to 0, differences in inference frameworks, batch sizes, and hardware floating-point operations can all cause output drift.85% similarVerifiedEven when temperature is set to 0, some models still exhibit slight output fluctuations due to floating-point errors in parallel computation.81% similarUnverified"Temperature 0" is not equivalent to "a unique result": batch scheduling, floating-point nondeterminism, mixture-of-experts routing, quantization error, and provider backend redeployments can all cause the same input to produce different outputs.79% similarUnverifiedTemperature=0（贪心采样）也不能保证模型输出的确定性，至少有三个原因：浮点运算非结合性、MoE路由不确定性、服务端配置漂移79% similarUnverifiedSetting the sampling temperature to 0 causes reasoning models to fall into a thinking loop and become stuck during the thinking phase.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/526461API
curl https://kongchang.com/api/v1/knowledge/claims/526461MCP
get_claim(id=526461)