模型排行

排名	变化	模型	得分	置信区间
1	—	Sonic 3.6闭源Cartesia	1,283	±17
2	—	Qwen-Audio-3.0-TTS-Plus闭源Alibaba	1,240	±14
3	—	Simba 3.2闭源SpeechifyAI	1,240	±14
4	—	Luna TTS闭源VUI Labs	1,219	±14
5	—	Gemini 3.1 Flash TTS闭源Google	1,211	±12
6	—	StepAudio 2.5 TTS闭源StepFun	1,206	±14
7	—	Sonic 3.5闭源Cartesia	1,204	±13
8	—	Realtime TTS 1.5 Max闭源Inworld	1,195	±13
9	—	Lightning V3.1 Pro (Jul 2026)闭源Smallest.ai	1,193	±15
10	—	Realtime TTS-2 - Research Preview闭源Inworld	1,188	±13
11	—	Eleven v3闭源ElevenLabs	1,179	±11
12	—	Speech 2.8 HD闭源MiniMax	1,174	±11
13	—	Falcon 2闭源Murf AI	1,155	±15
14	—	Speech 2.8 Turbo闭源MiniMax	1,154	±11
15	—	Gradium TTS Beta闭源Gradium	1,151	±15
16	—	Async Flash v1.5闭源Async	1,150	±13
17	—	Fish Audio S2.1 Pro闭源Fish Audio	1,144	±13
18	—	Step TTS 2 (Mar 2026)闭源StepFun	1,141	±13
19	—	Speech 2.6 HD闭源MiniMax	1,136	±11
20	—	Lightning V3.1 Pro TTS (Jun 2026)闭源Smallest.ai	1,136	±13
21	—	Realtime TTS 1.5 Mini闭源Inworld	1,132	±12
22	—	Azure HD 2.5闭源Microsoft	1,132	±12
23	—	Async Pro v1.0闭源Async	1,129	±13
24	—	Speech 2.6 Turbo闭源MiniMax	1,127	±11
25	—	Fish Audio S2 Pro开源Fish Audio	1,126	±13
26	—	Simba 3.0闭源SpeechifyAI	1,124	±12
27	—	SpaceXAI TTS闭源SpaceXAI	1,118	±16
28	—	Speech-02-HD闭源MiniMax	1,111	±11
29	—	TTS-1 HD闭源OpenAI	1,107	±12
30	—	Step Audio EditX (Mar 2026)开源StepFun	1,107	±13
31	—	Turbo v2.5闭源ElevenLabs	1,104	±10
32	—	Multilingual v2闭源ElevenLabs	1,104	±10
33	—	Chatterbox HD闭源Resemble AI	1,095	±13
34	—	TTS-1闭源OpenAI	1,094	±11
35	—	Gradium TTS闭源Gradium	1,093	±13
36	—	Gemini 2.5 Flash Lite TTS闭源Google	1,089	±12
37	—	Speech-02-Turbo闭源MiniMax	1,085	±11
38	—	Flash v2.5闭源ElevenLabs	1,084	±11
39	—	Studio闭源Google	1,081	±11
40	—	Voxtral TTS开源Mistral	1,079	±13
41	—	Speech 2.6闭源Hithink	1,077	±13
42	—	OpenAudio S1闭源Fish Audio	1,076	±11
43	—	GPT-Realtime-2闭源OpenAI	1,072	±13
44	—	Sonic 3闭源Cartesia	1,072	±11
45	—	Bland Speech v3闭源Bland AI	1,071	±12
46	—	Maya 2 Flash闭源Maya Research	1,070	±14
47	—	Journey闭源Google	1,069	±13
48	—	T2A-01-HD闭源MiniMax	1,065	±11
49	—	Magpie-Multilingual 357M (Feb 2026)开源NVIDIA	1,065	±13
50	—	SIMBA 1.6闭源SpeechifyAI	1,065	±12
51	—	Polly Generative闭源Amazon	1,064	±12
52	—	MiMo-V2.5-TTS闭源Xiaomi	1,061	±13
53	—	Kokoro 82M v1.0开源Kokoro	1,059	±11
54	—	Gemini 2.5 Flash TTS (Dec 2025)闭源Google	1,058	±12
55	—	Coda闭源Rime	1,058	±13
56	—	Chirp 3: HD闭源Google	1,057	±12
57	—	Octave 2闭源Hume AI	1,057	±12
58	—	Maya 2 Global闭源Maya Research	1,052	±14
59	—	Async Flash v1.0闭源Async	1,051	±11
60	—	OpenAudio S1 Mini闭源Fish Audio	1,050	±19
61	—	Maya1开源Maya Research	1,046	±12
62	—	Sonic English (Oct 2024)闭源Cartesia	1,044	±12
63	—	Polly Long-Form闭源Amazon	1,042	±13
64	—	Higgs Audio V3 TTS开源Boson AI	1,042	±13
65	—	Gemini 2.5 Pro (Dec 2025)闭源Google	1,039	±12
66	—	SIMBA 1.0闭源SpeechifyAI	1,037	±11
67	—	T2A-01-Turbo闭源MiniMax	1,032	±11
68	—	Azure Neural闭源Microsoft	1,031	±19
69	—	Octave TTS闭源Hume AI	1,031	±11
70	—	MAI-Voice-1闭源Microsoft	1,029	±13
71	—	Lightning v3.1闭源Smallest.ai	1,023	±13
72	—	Chatterbox开源Resemble AI	1,020	±11
73	—	MiMo-V2-TTS闭源Xiaomi	1,019	±13
74	—	Magpie-Multilingual 357M开源NVIDIA	1,005	±12
75	—	Arcana v3闭源Rime	1,004	±13
76	—	Zonos-v0.1开源Zyphra	1,000	±0
77	—	Murf Speech Gen 2闭源Murf AI	981	±11
78	—	LMNT闭源LMNT	977	±12
79	—	VibeVoice 1.5B闭源Microsoft	971	±13
80	—	VibeVoice 7B开源Microsoft	966	±13
81	—	OpenVoice v2开源OpenVoice	955	±13
82	—	Magpie Multilingual闭源NVIDIA	941	±14
83	—	Qwen3 TTS Flash闭源Alibaba	940	±14
84	—	Neuphonic TTS闭源Neuphonic	936	±14
85	—	Qwen3 TTS闭源Alibaba	925	±13
86	—	XTTS v2开源Coqui	921	±14
87	—	WaveNet闭源Google	913	±12
88	—	Mist V2闭源Rime	899	±14
89	—	StyleTTS 2开源StyleTTS	892	±15
90	—	Neural2闭源Google	890	±12
91	—	Polly Neural闭源Amazon	887	±14
92	—	Standard闭源Google	880	±12
93	—	Noiz TTS闭源Noiz	865	±15
94	—	MetaVoice v1开源MetaVoice	845	±17
95	—	Polly Standard闭源Amazon	817	±15

其余 45 个模型

智力进步

每家厂商当时最强的模型在 Artificial Analysis 智力指数上的分数，按模型发布的月份画。

涨得最多 · Anthropic +57.2目前最高 · Anthropic 64.6最高与最低相差 18.1 分时间范围 2025年11月 – 2026年11月

64.6Anthropic60.0Google56.9OpenAI50.1Meta46.5Kimi