annex1-002

GMP KnowledgeEu Annex1medium

Question

Under the revised EU GMP Annex 1, what is the recommended maximum time limit for holding filtered bulk solution before filling in an aseptic process, unless otherwise justified?

Choices

A. No specific time limit is mentioned
B. The holding time should be minimised and a time limit established during process validation
C. 48 hours at controlled room temperature
D. 24 hours under refrigerated conditions only

Cross-Model Comparison

Model	Score	Latency	Tokens In	Tokens Out
Qwen3.8-Flash	N/A	N/A	-	-
GLM-5.3	N/A	N/A	-	-
MiniMax M3	N/A	N/A	-	-
Hy4 Preview	N/A	N/A	-	-
Muse Glimmer 30B	N/A	N/A	-	-
Qwen3.8 Max (2.4T-A95B)	N/A	N/A	-	-
DeepSeek-R1-Distill-Qwen-32B	100.0%	24.8s	88	510
Llama 4 Scout	100.0%	4.8s	92	383
Llama 4 Maverick	100.0%	14.1s	92	327
Mistral Small 2603	100.0%	1.2s	100	172
DeepSeek-V3.2	100.0%	5.6s	87	134
Qwen3.5-397B-A17B	100.0%	39.2s	96	2,153
DeepSeek-R1	100.0%	10.2s	93	334
Mistral Large 3 675B	100.0%	4.8s	88	278
Qwen3.5-35B-A3B	100.0%	54.9s	96	7,284
Gemini 3.1 Pro	100.0%	9.7s	83	990
Gemini 3 Flash	100.0%	3.0s	83	274
Gemini 3.1 Flash-Lite	100.0%	1.9s	83	145
Qwen3.6 27B	100.0%	24.1s	96	1,577
Qwen3.6 35B A3B	100.0%	6.9s	96	1,308
DeepSeek V4 Pro	100.0%	10.3s	87	399
DeepSeek-V3.2	100.0%	7.3s	93	160
Gemma 4 26B A4B IT	100.0%	3.8s	96	272
Gemma 4 31B IT	100.0%	6.0s	96	266
MiniMax M2.7	100.0%	8.5s	119	483
GLM-5.2	100.0%	14.7s	94	478
Llama 4 Maverick	100.0%	17.5s	150	372
Claude Opus 4.8	100.0%	4.4s	137	275
Llama 4 Scout	100.0%	8.8s	150	344
Llama 3.3 70B Instruct	100.0%	6.8s	153	136
Mistral Small 2603	100.0%	2.4s	163	298
Mistral Large 3 675B	100.0%	3.5s	151	263
Qwen3.5-35B-A3B	100.0%	13.9s	158	3,339
Qwen3.6 27B	100.0%	153.6s	158	5,738
Qwen3.6 35B A3B	100.0%	16.2s	158	1,805
Qwen3.5-397B-A17B	100.0%	28.4s	158	2,692
DeepSeek V4 Pro	100.0%	22.7s	147	50
DeepSeek-V3.2	100.0%	3.8s	147	75
DeepSeek V4 Flash	100.0%	13.1s	147	66
DeepSeek-R1	100.0%	72.0s	146	1,191
Gemma 4 26B A4B IT	100.0%	7.6s	158	308
Gemma 4 31B IT	100.0%	14.3s	158	168
MiniMax M2.7	100.0%	10.0s	182	647
GLM-5.2	100.0%	172.7s	154	6,822
Qwen3.8 27B	100.0%	11.8s	268	1,571
Muse Glimmer 30B	100.0%	17.4s	197	1,076
Kimi K3	100.0%	380.5s	227	9,573
GPT-5.4	100.0%	1.6s	86	86
GLM-5.3-Flash	100.0%	300.7s	154	10,683
GPT-5.4 mini	100.0%	939ms	86	73
GPT-5.4 nano	100.0%	673ms	86	34
Claude Haiku 4.5	100.0%	3.8s	96	320
Claude Sonnet 4.6	100.0%	9.8s	96	349
Claude Opus 4.6	100.0%	12.7s	96	437
Claude Haiku 4.5	100.0%	3.8s	160	304
GPT-5.6 Sol	100.0%	3.3s	144	124
GPT-5.6 Terra	100.0%	4.2s	144	151
GPT-5.6 Luna	100.0%	3.0s	144	175
Hy4 Preview	100.0%	125.4s	164	10,050
MiniMax M3	100.0%	13.0s	317	830
GLM-5.3	100.0%	33.2s	154	1,328
Claude Opus 5	100.0%	6.5s	225	299
Claude Sonnet 5	100.0%	1.0s	225	9
DeepSeek V4 Flash	0.0%	39.6s	87	4,000
Qwen3.8-Flash	0.0%	156.2s	209	12,992
Llama 3.3 70B Instruct	0.0%	119.2s	94	4,000
Qwen3.8 Max (2.4T-A95B)	0.0%	161.0s	209	6,925

annex1-002

Cross-Model Comparison

Tags