By folio

Scores out of 100, one per folio
Circled: under 50; failed: timeout, token limit or request error
Highlighted: 1st and 2nd on each folio by character error
Median: middle score among models that ran the folio
Select a score to open its transcription, a heading to sort

Model	Colour images					Microfilm
Model
Gemini 4 Argon (multi-step)	99	99	96	96	99	99	99	98
Gemini 4 Argon (high)	99	99	94	96	99	99	99	98
Gemini 4 Argon (low)	99	98	94	96	99	99	96	97
Gemini 3 Flash (low)	99	99	89	93	99	88	97	95
Gemini 3 Pro (high)	98	98	91	94	96	92	95	95
Gemini 3.8 Flash (low)	98	97	90	93	98	94	94	95
Opus 5.5 (high)	97	97	91	94	99	90	94	95
Gemini 3.5 Flash (low)	98	99	87	92	97	92	96	94
Gemini 3 Pro (low)	98	97	90	93	99	84	94	94
Gemini 3.5 Flash (high)	98	96	90	94	97	85	95	94
Gemini 3 Flash (high)	98	93	90	92	97	86	92	93
Gemini 3.1 Pro (low)	96	96	77	92	99	89	95	92
Gemini 3.1 Pro (high)	96	94	77	92	99	82	94	91
Opus 5.5 (low)	99	98	87	95	98	89	65	90
GPT-6.1 Sol (high)	98	94	failed	93	97	failed	9	785 of 7 run
Sonnet 5.5 (low)	97	97	82	91	96	36	23	75
Opus 4.6 (thinking)	88	96	66	79	94	51	29	72
Sonnet 5.5 (high)	97	98	46	91	96	55	18	72
Opus 4.7	97	97	61	92	94	27	31	71
Muse Spark Open	95	98	54	87	90	34	38	71
Opus 4.7 (thinking)	98	97	60	92	92	23	29	70
Opus 4.6	91	88	61	79	95	42	32	70
GPT-6.1 Sol (low)	98	98	64	92	93	31	6	69
Sonnet 4.6	84	93	56	80	83	35	27	65
Kimi K3 (no thinking)	90	97	62	83	88	34	1	65
Kimi K3	96	97	48	failed	94	32	15	646 of 7 run
Kimi K2.6	failed	95	48	86	91	28	28	636 of 7 run
Haiku 5.5 (low)	88	95	24	84	80	28	31	61
Haiku 5.5 (high)	92	95	32	80	85	16	27	61
Sonnet 4.6 (thinking)	71	92	55	66	83	32	28	61
DeepSeek V4.1 Flash 1017	94	95	failed	37	52	failed	12	585 of 7 run
Gemini 3.8 Flash (high)	17	17	91	91	97	1	89	58
GPT-6 Astra (low)	98	94	0	95	98	0	0	55
Qwen 3.6 Plus	85	95	0	74	75	23	29	54
GPT-5.2 (low)	89	98	28	70	85	0	0	53
GPT-5.2 (high)	92	96	49	35	76	8	3	51
GPT-5.5 (low)	22	96	50	53	25	79	0	46
DeepSeek V4.1 Flash 1017 (no thinking)	85	93	18	22	75	26	0	46
Qwen 3.6 35B A3B	40	85	22	52	57	26	28	44
Grok 4.3	59	80	39	49	51	11	0	41
GPT-4.1	43	94	32	43	34	26	14	41
GPT-5.4 Mini (low)	47	86	45	51	27	27	0	40
Qwen 3.8 27B	39	77	8	52	56	26	24	40
GPT-5 Mini (low)	37	50	42	48	52	27	3	37
Haiku 4.5	48	49	45	35	43	28	7	36
GPT-5 Mini (high)	48	51	42	40	53	0	0	33
Haiku 4.5 (thinking)	49	50	40	33	22	28	6	33
Llama 4 Maverick	31	37	37	0	45	29	27	29
Llama 4 Scout	31	29	37	26	44	28	10	29
Nemotron 3.5 Super VL	failed	7	failed	44	25	failed	failed	253 of 7 run
GLM 5.3 Flash	71	89	1	3	4	1	1	24
GPT-5.4 Mini (high)	44	91	0	17	8	4	0	23
Nemotron 3.5 Super VL (no thinking)	0	36	0	0	0	0	0	5
GPT-6 Luna (low)	7	0	1	0	11	2	0	3
GPT-6 Luna (high)	2	0	0	0	0	5	0	1
Median	91	95	50	80	88	29	27	61