Best LLM

Which LLM understands images best?

Claude Opus 5.5 and Claude Fable 5 are practically level at the top in image understanding.

Among the ten best, Gemini 3.8 Flash is the cheapest at $1.50 per million tokens.

The fastest among them is Muse Spark 1.3, at 102 tokens per second.

Ranking: Image understanding

Place and value in the area, plus overall index, price and speed. Higher values are better.

Rank	Model	Value	Overall index	Price (3:1)	Speed
1	Claude Opus 5.5Top group	124.3	125.6	$8	75
2	Claude Fable 5.1Upper group	121.1	124.1	$20	59
3	Claude Fable 5Upper group	119.6	112.8	$20	50
4	GPT-6 AstraUpper group	118.6	120.2	$20	58
5	GPT-6.1 SolUpper group	118.4	118.0	$4	44
6	Claude Opus 5Upper group	115.5	117.4	$10	71
7	Gemini 3.8 FlashMiddle group	112.3	106.8	$1.50	81
8	Gemini 3.7 FlashMiddle group	112.3	107.9	n/a	72
9	GPT-5.6 SolMiddle group	111.4	109.1	$8	75
10	Muse Spark 1.3Middle group	109.4	107.3	$2	102
11	Gemini 3.5 FlashMiddle group	104.3	97.8	n/a	142
12	GPT-6 SolMiddle group	103.6	109.1	$4	82
13	Gemini 3.6 FlashMiddle group	103.5	97.6	$1.50	110
14	GPT-5.6 TerraMiddle group	99.3	100.8	$4.50	72
15	Gemini 3.1 Pro PreviewMiddle group	98.4	91.4	$4.50	96
16	GLM-5.3-FlashMiddle group	97.2	94.3	$0.24	101
17	DeepSeek-V4.1-FlashMiddle group	94.8	99.5	$0.26	118
18	Grok 4.7Middle group	93.3	103.3	$3	47
19	GPT-5.6 LunaMiddle group	91.3	93.8	$0.45	123
20	Kimi K2.6Middle group	91.2	84.9	$1.71	32

Show all 47 ranked models

Rank	Model	Value	Overall index	Price (3:1)	Speed
21	Claude Sonnet 5Middle group	89.2	92.1	$4	77
22	GPT-5.4Middle group	88.9	86.5	$5.63	58
23	Claude Sonnet 4.6Middle group	88.8	86.7	$6	38
24	GPT-6 LunaMiddle group	88.4	95.8	$0.20	77
25	Gemini 3.5 Flash-LiteLower group	84.1	73.7	$0.85	145
26	Grok 4.5Lower group	82.1	94.8	$3	52
27	MiniMax M3Lower group	78.6	83.0	$0.52	82
28	Claude Opus 4.6Lower group	78.1	84.2	$10	36
29	Grok 4.3Lower group	76.6	71.6	$1.56	66
30	GPT-5.4 miniLower group	75.2	75.1	$1.69	134
31	Gemini 2.5 ProLower group	73.5	61.4	$3.44	74
32	MiMo-V2.5Lower group	71.5	69.1	$0.18	55
33	GPT-5Lower group	70.1	77.5	$3.44	83
34	o3Lower group	69.9	74.2	$3.50	73
35	o4-miniLower group	66.3	63.5	n/a	n/a
36	Inkling SmallLower group	63.8	68.7	$0.52	61
37	GPT-5 miniLower group	59.6	65.8	$0.69	105
38	GPT-5.4 nanoLower group	53.0	64.0	$0.46	98
39	Gemini 2.5 FlashLower group	49.6	n/a	$0.85	n/a
40	Mistral Medium 3.5Lower group	46.4	46.3	$3	118
41	GPT-4.1Lower group	43.8	43.6	$3.50	81
42	GPT-4.1 miniLower group	33.0	32.3	$0.70	95
43	Mistral Large 3Lower group	32.9	32.7	$0.75	86
44	GPT-5 nanoLower group	27.1	32.4	$0.14	136
45	Llama 4 MaverickLower group	15.3	5.1	$0.51	75
46	Llama 4 ScoutLower group	8.1	4.7	$0.15	62
47	GPT-4.1 nanoLower group	-19.8	-15.8	n/a	110

Price in dollars per million tokens, mixed from 3 parts input and 1 part output. Speed in tokens per second.

The top 20: Image understandingValue in the area · higher is better

Claude Opus 5.5Rank 1124.3

Claude Fable 5.1Rank 2121.1

Claude Fable 5Rank 3119.6

GPT-6 AstraRank 4118.6

GPT-6.1 SolRank 5118.4

Claude Opus 5Rank 6115.5

Gemini 3.8 FlashRank 7112.3

Gemini 3.7 FlashRank 8112.3

GPT-5.6 SolRank 9111.4

Muse Spark 1.3Rank 10109.4

Gemini 3.5 FlashRank 11104.3

GPT-6 SolRank 12103.6

Gemini 3.6 FlashRank 13103.5

GPT-5.6 TerraRank 1499.3

Gemini 3.1 Pro PreviewRank 1598.4

GLM-5.3-FlashRank 1697.2

DeepSeek-V4.1-FlashRank 1794.8

Grok 4.7Rank 1893.3

GPT-5.6 LunaRank 1991.3

Kimi K2.6Rank 2091.2

4 tests by independent evaluators feed this area. Each test counts with the result of the model in question.