Name: Claude Haiku 5.5
Price: 0.1 USD
Author: Anthropic

Rank	Model	Overall score	Artificial Analysis	LiveBench	Epoch AI	ARC-AGI	Scale SEAL	Vals AI	Terminal-Bench	EQ-Bench	SimpleBench	BenchLM
1	Claude Opus 5.5 Anthropic	92.7	#1	#2	#1	#3	—	#2	#1	—	#1	#1
2	Gemini 4 Argon Google	92.4	#5	—	—	—	—	#1	—	—	—	—
3	GPT-6 Astra OpenAI	91.3	#4	#4	#2	#1	#1	#6	#3	—	#4	#2
4	Claude Fable 5.1 Anthropic	90.9	#3	#1	#5	#6	#2	#4	#5	—	#2	#4
5	Claude Sonnet 5.5 Anthropic	90.8	#2	#20	#4	—	—	#2	#2	—	#14	#3
6	GPT-6.1 Sol OpenAI	90.2	#6	#5	#3	#2	—	#8	#3	—	#5	—
7	Claude Opus 5 Anthropic	89.8	#7	#10	#6	#5	—	#5	#6	#1	#9	#5
8	Claude Fable 5 Anthropic	89.4	#8	#3	#9	#8	—	#7	#8	#2	—	#6
9	GPT-6 Sol OpenAI	86.7	#10	#12	#7	#7	—	#11	#7	—	#17	#8
10	GPT-5.6 Sol OpenAI	86.6	#11	#7	#10	#4	—	#10	#11	#9	#25	#7
11	GPT-5.5 Pro OpenAI	85.7	—	—	#8	#11	—	—	—	—	#11	—
12	Muse Spark 1.3 Meta	85.7	#9	#5	#19	—	—	#9	#20	—	#8	—
13	GPT-5.4 Pro OpenAI	84.2	—	—	#13	#15	—	—	—	—	—	—
14	MiMo V2.6 Pro Xiaomi	84.1	#13	—	—	—	—	#12	—	—	—	—
15	GPT-5.5 OpenAI	82.7	#30	#9	#12	#10	—	—	—	#4	#21	#17
16	Gemini 3.8 Flash Google	82.3	#24	#32	#20	#8	#4	#15	#17	—	#6	#9
17	Grok 4.7 xAI	81.8	#12	#21	#41	#26	—	#14	#10	—	—	—
18	GLM-5.3 Z.ai (Zhipu)	81.0	#15	#30	#27	—	—	#16	#9	—	#24	#14
19	DeepSeek V4.1 Flash DeepSeek	80.8	#27	#7	#32	#19	—	#22	—	—	#23	#16
20	GPT-5.6 Terra OpenAI	80.4	#20	#19	#11	#14	—	#18	#15	#11	#55	#10
21	Kimi K3 Moonshot AI	80.3	#18	#12	#15	#29	—	#26	—	#3	#35	#11
22	GPT-5.4 OpenAI	80.2	—	#16	#17	#18	—	—	—	#7	—	#19
23	Claude Opus 4.8 Anthropic	80.2	#21	#28	#14	#20	—	#13	#14	#6	#25	#13
24	Grok 4.6 xAI	80.1	#16	#16	#21	#23	—	#19	#16	—	#14	#15
25	Qwen 3.8 Max Alibaba (Qwen)	80.0	#14	#15	#31	—	—	#30	#13	—	—	#12
26	Gemini 3.7 Flash Google	79.7	#28	#14	#16	#11	—	#22	#23	—	—	#18
27	MiMo V2.6 Flash Xiaomi	79.1	#34	—	—	—	—	#17	—	—	—	—
28	Step 5 StepFun	78.8	#17	—	—	—	—	#28	—	—	—	—
29	Claude Opus 4.7 Anthropic	78.3	—	#27	#25	#17	#9	—	—	#5	#28	—
30	Muse Spark 1.2 Meta	78.3	#26	#16	#33	—	—	#29	—	—	#16	#20
31	Qwen 3.8 Flash Next Alibaba (Qwen)	76.7	#25	#28	—	—	—	—	—	—	—	—
32	Claude Sonnet 5 Anthropic	75.6	#32	#31	#26	—	—	#20	#21	#10	#36	#23
33	GLM-5.3 Flash Z.ai (Zhipu)	74.4	#21	#51	#47	#24	—	—	#12	—	—	—
34	Grok 4.20 xAI	74.0	—	—	#46	#25	—	—	—	—	—	—
35	GPT-6 Luna OpenAI	73.3	#33	#48	#24	#34	—	#24	#19	—	—	—
36	DeepSeek V4 Pro DeepSeek	73.3	#36	#21	#29	#28	—	#33	—	#18	#53	—
37	Claude Opus 4.6 Anthropic	73.0	—	#38	#30	#22	#10	—	—	#13	#22	#28
38	Muse Spark 1.1 Meta	72.7	#41	#34	#38	—	—	—	—	#8	—	#22
39	GPT-5.6 Luna OpenAI	72.5	#35	#41	#23	#33	—	#21	#18	#19	#57	#21
40	Muse Spark Meta	72.5	—	—	#45	—	#6	—	—	—	—	#32
41	DeepSeek V4 Flash DeepSeek	72.4	#38	#39	#35	#26	—	#31	—	—	#33	—
42	Grok 4.5 xAI	72.3	#29	#32	#39	#37	—	#35	#21	—	#20	#24
43	Gemini 3.1 Pro Google	72.1	#47	#24	#34	#16	#3	#38	—	#21	#10	—
44	Claude Haiku 5.5 Anthropic	71.8	#19	#47	—	—	—	—	—	—	—	—
45	Mistral Large 4 Mistral AI	71.3	#30	#50	—	—	—	#31	—	—	—	—
46	Gemini 3.5 Flash Google	71.2	#44	#36	#36	#20	—	#34	—	#24	#12	#27
47	Gemini 3.6 Flash Google	69.9	#39	#41	#37	#29	—	—	—	—	—	#25
48	GPT-5.2 Pro OpenAI	68.0	—	—	#28	#35	—	—	—	—	#43	—
49	Gemini 3 Pro Google	67.5	—	—	#43	#43	#7	—	—	—	#13	#33
50	GPT-5.3 Codex OpenAI	67.5	—	—	#18	—	—	—	—	—	—	#30

Rank	Provider	Overall score	Top model	Models
1	Anthropic	87.4	Claude Opus 5.5	29
2	OpenAI	85.4	GPT-6 Astra	54
3	Google	77.7	Gemini 4 Argon	34
4	Meta	72.1	Muse Spark 1.3	23
5	xAI	70.0	Grok 4.7	13
6	Moonshot AI	68.2	Kimi K3	5
7	DeepSeek	68.1	DeepSeek V4.1 Flash	22
8	Alibaba (Qwen)	67.4	Qwen 3.8 Max	46
9	Z.ai (Zhipu)	65.7	GLM-5.3	10
10	StepFun	63.0	Step 5	1
11	Xiaomi	59.2	MiMo V2.6 Pro	3
12	Thinking Machines	40.9	Inkling	2
13	Tencent	39.9	—	2
14	MiniMax	36.3	Minimax M3	5
15	Mistral AI	35.6	Mistral Large 4	27
16	NVIDIA	29.2	Nemotron 3 Ultra 550B A55B	10
17	Amazon	18.3	—	3
18	Microsoft	13.4	Phi-4	6
19	Cohere	11.5	Command R	3
20	Inception	8.8	Mercury 2.5	1

A provider’s result on each site is that of its best model there.

Performance vs price

Top left = cheap and strong. Click a point for details.

Best value

Overall score ÷ price

Top open-weight models

Models with public weights

Fastest models

Output tokens per second

By country

Top 50 models

Agreement between leaderboards

Rank correlation on shared models (1 = identical).

How the ranking works

Collect — Official data from the 11 leaderboards below is fetched every hour. Scores marked as estimates are not used.
Match — Different spellings of the same model are merged. For variants such as reasoning effort, each site’s best result is used.
Standardize — Each site’s main metric is standardized within that site, keeping score gaps rather than just ranks.
Equate — A factor model, bridged by models listed on several sites, corrects for each site’s difficulty and model mix. Noisier sites get less weight.
No penalty for missing data — Models are scored only from the sites that rated them. Uncertainty is shown as a 90% interval. Models rated on 2+ sites are ranked.
Arena — Arena (formerly LMArena) is used for domain rankings only, not for the overall score. New models are added automatically.

Claude Haiku 5.5: benchmarks and ranking

Rank on each leaderboard

Domain scores