Cortex Lab

LLM performance leaderboard

Price, output speed and quality across every major model, compared side by side. This is the table we open before deciding what a workflow should run on. The data is published by Artificial Analysis — we host it here for convenience and credit it in full.

			Quality index			Cost $/1M tokens			Speed			Benchmarks
#	Model
1	Claude Opus 5 (Adaptive Reasoning, Max Effort) Anthropic	Jul 2026	63.1	78.0	—	$10.00	$5.00	$25.00	54	34.57	34.57	93.2%	54.9%	55.7%	75.7%	—	—	—	—	—	—	—	89.1%	—	42.1%
2	Claude Opus 5 (Adaptive Reasoning, Xhigh Effort) Anthropic	Jul 2026	62.5	77.0	—	$10.00	$5.00	$25.00	54	18.21	18.21	93.7%	54.4%	55.0%	76.3%	—	—	—	—	—	—	—	88.0%	—	43.3%
3	Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) Anthropic	Jun 2026	62.1	76.5	—	$20.00	$10.00	$50.00	62	54.13	54.13	92.6%	55.5%	60.2%	76.7%	63.5%	98.5%	62.9%	—	—	—	—	84.6%	—	38.1%
4	Claude Opus 5 (Adaptive Reasoning, High Effort) Anthropic	Jul 2026	61.5	76.5	—	$10.00	$5.00	$25.00	53	11.33	11.33	93.7%	52.8%	54.3%	76.3%	—	—	—	—	—	—	—	87.6%	—	44.7%
5	GPT-5.6 Sol (max) OpenAI	Jul 2026	60.9	77.4	—	$8.00	$4.00	$20.00	84	50.56	50.56	94.1%	49.5%	56.1%	77.7%	72.7%	85.1%	65.9%	—	—	—	—	88.0%	—	44.3%
6	Grok 4.6 (high) SpaceXAI	Aug 2026	60.9	76.8	—	$3.00	$2.00	$6.00	58	36.59	36.59	94.9%	42.9%	53.6%	75.0%	—	—	—	—	—	—	—	88.4%	—	50.7%
7	Grok 4.6 (xhigh) SpaceXAI	Aug 2026	60.0	75.9	—	$3.00	$2.00	$6.00	60	38.68	38.68	93.5%	44.1%	51.6%	75.7%	—	—	—	—	—	—	—	88.0%	—	43.3%
8	Kimi K3 (max) Kimi	Jul 2026	59.7	76.2	—	$6.00	$3.00	$15.00	36	3.56	58.53	93.5%	46.9%	58.7%	82.7%	—	—	—	—	—	—	—	85.0%	—	46.0%
9	GLM-5.3 (max) Z AI	Aug 2026	59.5	74.8	—	$2.15	$1.40	$4.40	75	1.61	28.20	91.7%	42.3%	56.5%	76.3%	—	—	—	—	—	—	—	83.9%	—	50.3%
10	GPT-5.6 Sol (xhigh) OpenAI	Jul 2026	59.0	78.3	—	$8.00	$4.00	$20.00	82	23.49	23.49	93.1%	47.3%	56.0%	76.3%	71.0%	84.8%	61.4%	—	—	—	—	89.5%	—	38.1%
11	Grok 4.6 (medium) SpaceXAI	Aug 2026	59.0	74.4	—	$3.00	$2.00	$6.00	57	31.70	31.70	93.5%	42.1%	54.6%	72.7%	—	—	—	—	—	—	—	84.3%	—	44.3%
12	Claude Opus 5 (Adaptive Reasoning, Medium Effort) Anthropic	Jul 2026	58.6	74.3	—	$10.00	$5.00	$25.00	51	8.65	8.65	91.9%	51.3%	50.7%	78.7%	—	—	—	—	—	—	—	86.1%	—	38.6%
13	Qwen3.8 Max Alibaba	Aug 2026	58.1	71.8	—	$3.00	$2.00	$6.00	30	1.76	68.67	92.7%	43.0%	52.9%	74.3%	—	—	—	—	—	—	—	81.3%	—	51.3%
14	Qwen3.8 2.4T A95B Alibaba	Aug 2026	57.7	71.9	—	$3.00	$2.00	$6.00	31	1.91	65.63	93.5%	42.4%	51.6%	75.3%	—	—	—	—	—	—	—	82.0%	—	49.1%
15	GLM-5.3-Flash Z AI	Aug 2026	57.5	71.5	—	$0.24	$0.15	$0.50	50	1.11	41.02	91.2%	39.9%	46.1%	78.0%	—	—	—	—	—	—	—	84.3%	—	47.2%
16	GPT-5.6 Sol (high) OpenAI	Jul 2026	57.3	77.2	—	$8.00	$4.00	$20.00	78	14.11	14.11	92.8%	46.0%	56.9%	75.3%	69.2%	83.3%	62.1%	—	—	—	—	87.3%	—	36.7%
17	Claude Opus 4.8 (Adaptive Reasoning, Max Effort) Anthropic	May 2026	57.3	74.3	—	$10.00	$5.00	$25.00	0	0.00	0.00	92.0%	48.7%	53.5%	73.0%	62.2%	94.4%	58.3%	—	—	—	—	84.6%	—	34.2%
18	Muse Spark 1.2 (xhigh) Meta	Aug 2026	56.8	72.2	—	$2.00	$1.25	$4.25	0	0.00	0.00	90.4%	45.5%	56.4%	83.3%	—	—	—	—	—	—	—	80.1%	—	34.8%
19	GPT-5.6 Terra (max) OpenAI	Jul 2026	56.6	76.7	—	$4.50	$2.00	$12.00	110	119.68	119.68	92.5%	42.9%	53.9%	79.7%	71.2%	86.3%	57.6%	—	—	—	—	88.0%	—	40.2%
20	GPT-5.5 (xhigh) OpenAI	Apr 2026	56.3	74.9	—	$11.25	$5.00	$30.00	0	0.00	0.00	93.5%	45.8%	56.1%	79.0%	75.9%	93.9%	60.6%	—	—	—	—	84.3%	—	39.0%
21	Gemini 3.7 Flash (high) Google	Aug 2026	56.0	76.1	—	$1.50	$0.75	$3.75	345	7.57	7.57	94.5%	47.9%	56.8%	80.0%	—	—	—	—	—	—	—	85.8%	—	32.8%
22	Qwen3.8-Flash-Next Alibaba	Aug 2026	55.8	73.1	—	$0.23	$0.15	$0.47	83	1.22	25.36	92.3%	38.0%	46.9%	77.0%	—	—	—	—	—	—	—	86.1%	—	45.4%
23	Grok 4.5 (high) SpaceXAI	Jul 2026	55.8	72.4	—	$3.00	$2.00	$6.00	54	13.12	13.12	93.1%	42.7%	54.1%	74.0%	—	—	—	—	—	—	—	81.6%	—	42.1%
24	GPT-5.6 Sol (medium) OpenAI	Jul 2026	55.6	76.3	—	$8.00	$4.00	$20.00	78	5.35	5.35	92.6%	42.2%	56.5%	74.3%	69.6%	81.0%	62.9%	—	—	—	—	86.1%	—	36.5%
25	Claude Sonnet 5 (Adaptive Reasoning, Max Effort) Anthropic	Jun 2026	55.3	71.5	—	$4.00	$2.00	$10.00	91	121.77	121.77	91.1%	41.3%	53.6%	77.0%	—	—	—	—	—	—	—	80.5%	—	37.3%
26	Claude Opus 4.7 (Adaptive Reasoning, Max Effort) Anthropic	Apr 2026	55.0	73.6	—	$10.00	$5.00	$25.00	0	0.00	0.00	91.4%	42.3%	54.5%	75.3%	58.6%	88.6%	51.5%	—	—	—	—	83.1%	—	34.6%
27	GPT-5.5 (high) OpenAI	Apr 2026	54.7	71.6	—	$11.25	$5.00	$30.00	0	0.00	0.00	93.2%	45.0%	55.9%	79.0%	71.6%	93.0%	59.8%	—	—	—	—	79.4%	—	36.7%
28	Gemini 3.7 Flash (medium) Google	Aug 2026	53.4	71.5	—	$1.50	$0.75	$3.75	327	4.42	4.42	92.1%	39.0%	57.9%	81.0%	—	—	—	—	—	—	—	78.3%	—	35.5%
29	DeepSeek V4 Pro 0813 (Reasoning, Max Effort) DeepSeek	Aug 2026	53.2	68.8	—	$1.98	$1.32	$3.96	62	1.15	33.25	92.8%	41.0%	49.2%	75.3%	—	—	—	—	—	—	—	78.7%	—	39.6%
30	Muse Spark 1.1 (xhigh) Meta	Jul 2026	53.2	71.3	—	$2.00	$1.25	$4.25	0	0.00	0.00	89.8%	46.2%	58.2%	81.3%	—	—	—	—	—	—	—	77.9%	—	31.8%
31	GPT-5.4 (xhigh) OpenAI	Mar 2026	53.1	71.1	—	$5.63	$2.50	$15.00	0	0.00	0.00	92.0%	43.7%	56.6%	77.7%	73.9%	87.1%	57.6%	—	—	—	—	78.3%	—	39.6%
32	GPT-5.6 Terra (xhigh) OpenAI	Jul 2026	52.8	70.6	—	$4.50	$2.00	$12.00	100	15.63	15.63	90.8%	41.9%	51.6%	75.0%	66.3%	80.4%	62.9%	—	—	—	—	80.1%	—	29.7%
33	GLM-5.2 (max) Z AI	Jun 2026	52.6	68.8	—	$2.15	$1.40	$4.40	68	1.28	30.53	89.5%	41.1%	50.5%	76.7%	73.3%	99.1%	50.8%	—	—	—	—	77.9%	—	34.6%
34	Claude Opus 5 (Adaptive Reasoning, Low Effort) Anthropic	Jul 2026	52.5	66.9	—	$10.00	$5.00	$25.00	52	2.59	2.59	88.9%	43.4%	48.0%	77.0%	—	—	—	—	—	—	—	76.4%	—	30.3%
35	GPT-5.6 Luna (max) OpenAI	Jul 2026	52.3	71.4	—	$0.45	$0.20	$1.20	129	69.71	69.71	91.1%	39.5%	52.5%	78.3%	—	—	—	—	—	—	—	80.9%	—	31.1%
36	Qwen3.8 27B (xhigh) Alibaba	Aug 2026	52.0	68.1	—	$1.13	$0.50	$3.00	46	1.26	45.01	90.5%	33.9%	44.7%	77.3%	—	—	—	—	—	—	—	79.8%	—	48.0%
37	Gemini 3.5 Flash (high) Google	May 2026	52.0	70.1	—	$3.38	$1.50	$9.00	0	0.00	0.00	92.2%	42.7%	53.1%	81.0%	76.3%	95.3%	40.9%	—	—	—	—	78.7%	—	32.2%
38	DeepSeek V4 Flash 0731 (Reasoning, Max Effort) DeepSeek	Jul 2026	51.8	69.1	—	$0.66	$0.44	$1.32	135	0.84	15.61	90.8%	38.6%	49.9%	74.3%	—	—	—	—	—	—	—	78.7%	—	39.4%
39	Grok 4.6 (low) SpaceXAI	Aug 2026	51.7	66.3	—	$3.00	$2.00	$6.00	54	3.96	3.96	87.9%	27.6%	48.4%	78.7%	—	—	—	—	—	—	—	75.3%	—	38.1%
40	Gemini 3.6 Flash (high) Google	Jul 2026	51.6	69.2	—	$1.50	$0.75	$3.75	175	12.41	12.41	92.8%	40.8%	52.7%	79.0%	—	—	—	—	—	—	—	77.5%	—	29.9%
41	DeepSeek V4 Flash Vision (Reasoning, Max Effort) DeepSeek	Aug 2026	51.5	65.0	—	$0.66	$0.44	$1.32	115	0.97	18.41	91.3%	34.5%	46.6%	78.0%	—	—	—	—	—	—	—	74.2%	—	41.0%
42	GPT-5.5 (medium) OpenAI	Apr 2026	51.4	71.5	—	$11.25	$5.00	$30.00	0	0.00	0.00	92.6%	42.4%	53.5%	76.7%	71.0%	91.8%	57.6%	—	—	—	—	80.5%	—	29.9%
43	Gemini 3.7 Flash (low) Google	Aug 2026	50.9	71.0	—	$1.50	$0.75	$3.75	329	0.87	0.87	90.1%	35.1%	53.6%	78.3%	—	—	—	—	—	—	—	79.8%	—	29.5%
44	GPT-5.6 Sol (low) OpenAI	Jul 2026	50.7	69.7	—	$8.00	$4.00	$20.00	76	2.46	2.46	89.8%	39.4%	55.4%	73.0%	66.5%	76.0%	60.6%	—	—	—	—	76.8%	—	29.1%
45	GPT-5.6 Luna (xhigh) OpenAI	Jul 2026	50.1	68.6	—	$0.45	$0.20	$1.20	122	35.27	35.27	89.5%	37.0%	50.0%	73.3%	—	—	—	—	—	—	—	77.9%	—	28.7%
46	GPT-5.6 Terra (high) OpenAI	Jul 2026	50.1	67.1	—	$4.50	$2.00	$12.00	96	2.56	2.56	89.6%	38.5%	50.1%	73.3%	64.4%	78.4%	57.6%	—	—	—	—	75.7%	—	28.7%
47	Agnes 2.5 Pro Beta Sapiens AI	Aug 2026	49.1	62.3	—	$0.15	$0.10	$0.30	151	1.44	14.68	90.5%	37.5%	47.6%	78.0%	—	—	—	—	—	—	—	69.7%	—	35.7%
48	Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort) Anthropic	Feb 2026	48.4	63.0	—	$6.00	$3.00	$15.00	0	0.00	0.00	87.5%	33.6%	46.8%	74.0%	56.6%	75.7%	53.0%	—	—	—	—	71.2%	—	34.4%
49	Kimi K3 (low) Kimi	Jul 2026	48.3	72.0	—	$6.00	$3.00	$15.00	37	3.51	57.82	84.2%	25.0%	51.2%	77.0%	—	—	—	—	—	—	—	82.4%	—	41.6%
50	Gemini 3.1 Pro Preview Google	Feb 2026	47.7	68.8	—	$4.50	$2.00	$12.00	120	22.68	22.68	94.1%	47.0%	58.9%	79.0%	77.1%	95.6%	53.8%	—	—	—	—	73.8%	—	21.4%
51	Motif 3 Motif Technologies	Aug 2026	47.4	63.5	—	$0.00	$0.00	$0.00	0	0.00	0.00	83.4%	37.0%	40.6%	72.3%	—	—	—	—	—	—	—	74.9%	—	35.3%
52	GPT-5.6 Luna (high) OpenAI	Jul 2026	47.0	63.3	—	$0.45	$0.20	$1.20	111	10.32	10.32	89.2%	33.4%	50.7%	74.0%	—	—	—	—	—	—	—	69.7%	—	25.2%
53	GPT-5.6 Terra (medium) OpenAI	Jul 2026	46.8	64.7	—	$4.50	$2.00	$12.00	97	1.59	1.59	87.2%	33.3%	49.7%	70.3%	62.2%	72.8%	—	—	—	—	—	72.3%	—	25.6%
54	Gemini 3.5 Flash (medium) Google	May 2026	46.7	—	—	$3.38	$1.50	$9.00	271	10.43	10.43	92.1%	41.3%	53.0%	79.7%	74.6%	95.6%	39.4%	—	—	—	—	—	—	—
55	Qwen3.7 Max Alibaba	May 2026	46.7	66.0	—	$3.75	$2.50	$7.50	0	0.00	0.00	92.3%	40.5%	48.8%	74.7%	80.5%	94.7%	50.8%	—	—	—	—	74.5%	—	11.8%
56	GPT-5.3 Codex (xhigh) OpenAI	Feb 2026	45.5	—	—	$4.81	$1.75	$14.00	126	36.41	36.41	91.5%	42.5%	53.2%	78.3%	75.4%	86.0%	53.0%	—	—	—	—	—	—	—
57	MiniMax-M3 MiniMax	Jun 2026	45.4	58.6	—	$0.53	$0.30	$1.20	126	0.98	16.86	92.9%	39.0%	45.4%	80.3%	82.9%	88.9%	42.4%	—	—	—	—	65.2%	—	15.3%
58	DeepSeek V4 Pro (Reasoning, Max Effort) DeepSeek	Apr 2026	45.3	59.4	—	$0.54	$0.44	$0.87	64	1.17	69.71	88.8%	37.5%	50.0%	70.0%	76.5%	96.2%	46.2%	—	—	—	—	64.0%	—	30.1%
59	Motif 3 (Beta) Motif Technologies	Jul 2026	45.3	62.0	—	$0.00	$0.00	$0.00	0	0.00	0.00	86.9%	40.4%	44.3%	70.7%	—	—	—	—	—	—	—	70.8%	—	—
60	Kimi K2.6 Kimi	Apr 2026	45.1	61.8	—	$1.71	$0.95	$4.00	0	0.00	0.00	91.1%	37.5%	53.5%	76.7%	76.0%	95.9%	43.9%	—	—	—	—	65.9%	—	23.3%
61	Claude Opus 4.6 (Adaptive Reasoning, Max Effort) Anthropic	Feb 2026	44.9	—	—	$10.00	$5.00	$25.00	0	0.00	0.00	89.6%	39.9%	51.9%	74.3%	53.1%	92.1%	46.2%	—	—	—	—	—	—	—
62	Qwen3.8 27B (medium) Alibaba	Aug 2026	44.5	56.1	—	$1.13	$0.50	$3.00	51	1.21	40.05	84.5%	14.1%	38.1%	76.3%	—	—	—	—	—	—	—	65.2%	—	47.4%
63	GPT-5.5 (low) OpenAI	Apr 2026	44.5	60.9	—	$11.25	$5.00	$30.00	0	0.00	0.00	91.0%	32.7%	51.6%	77.7%	64.4%	83.9%	52.3%	—	—	—	—	65.5%	—	24.9%
64	Muse Spark Meta	Apr 2026	44.3	58.6	—	$0.00	$0.00	$0.00	0	0.00	0.00	88.4%	40.7%	51.5%	77.0%	75.9%	91.5%	45.5%	—	—	—	—	62.2%	—	—
65	Claude Opus 4.7 (Non-reasoning, High Effort) Anthropic	Apr 2026	43.9	—	—	$10.00	$5.00	$25.00	0	0.00	0.00	88.5%	33.3%	50.1%	72.3%	43.6%	74.0%	54.5%	—	—	—	—	—	—	—
66	DeepSeek V4 Pro (Reasoning, High Effort) DeepSeek	Apr 2026	43.7	58.7	—	$0.54	$0.44	$0.87	66	1.22	31.21	90.5%	35.2%	46.4%	67.0%	71.3%	94.2%	41.7%	—	—	—	—	64.8%	—	26.2%
67	GPT-5.2 (xhigh) OpenAI	Dec 2025	43.3	—	99.0	$4.81	$1.75	$14.00	0	0.00	0.00	90.3%	37.7%	52.1%	79.3%	75.4%	84.8%	47.0%	87.4%	88.9%	99.0%	—	—	—	—
68	Kimi K2.7 Code Kimi	Jun 2026	43.0	60.8	—	$1.71	$0.95	$4.00	43	1.28	53.63	89.6%	35.0%	47.5%	75.0%	63.1%	90.1%	44.7%	—	—	—	—	67.4%	—	20.2%
69	MiMo-V2.5-Pro Xiaomi	Apr 2026	42.9	60.2	—	$0.54	$0.44	$0.87	37	4.56	58.10	86.6%	35.7%	50.2%	77.7%	79.9%	94.2%	43.2%	—	—	—	—	65.2%	—	9.9%
70	Qwen3.8 27B (low) Alibaba	Aug 2026	42.9	58.2	—	$1.13	$0.50	$3.00	51	1.38	40.89	84.5%	14.0%	39.8%	74.7%	—	—	—	—	—	—	—	67.4%	—	32.2%
71	Claude Sonnet 5 (Non-reasoning, High Effort) Anthropic	Jun 2026	42.6	66.4	—	$4.00	$2.00	$10.00	62	1.31	1.31	80.0%	19.0%	48.6%	66.0%	—	—	—	—	—	—	—	75.3%	—	15.7%
72	Inkling (xhigh) Thinking Machines	Jul 2026	42.3	52.1	—	$1.76	$1.00	$4.05	70	2.00	30.42	87.2%	31.9%	46.1%	73.3%	—	—	—	—	—	—	—	55.1%	—	29.1%
73	Hy3 Tencent	Jul 2026	42.2	58.8	—	$0.24	$0.14	$0.55	94	1.67	23.02	89.7%	33.5%	47.6%	74.7%	—	—	—	—	—	—	—	64.4%	—	22.9%
74	DeepSeek V4 Flash (Reasoning, Max Effort) DeepSeek	Apr 2026	42.1	56.2	—	$0.17	$0.13	$0.28	0	0.00	0.00	89.4%	34.8%	44.9%	70.0%	79.2%	95.0%	35.6%	—	—	—	—	61.8%	—	30.9%
75	GPT-5.6 Sol (Non-reasoning) OpenAI	Jul 2026	41.9	65.1	—	$8.00	$4.00	$20.00	71	1.18	1.18	79.0%	16.7%	47.1%	57.0%	—	—	—	—	—	—	—	74.2%	—	19.6%
76	Claude Opus 4.5 (Reasoning) Anthropic	Nov 2025	41.9	—	91.3	$10.00	$5.00	$25.00	0	0.00	0.00	86.6%	30.1%	49.5%	76.0%	58.0%	89.5%	47.0%	89.5%	87.1%	91.3%	—	—	—	—
77	Nex-N2-Pro Nex AGI	Jun 2026	41.7	59.1	—	$1.00	$0.50	$2.50	134	0.91	15.80	89.2%	33.7%	41.8%	76.3%	66.2%	81.6%	—	—	—	—	—	67.8%	—	17.5%
78	Solar Pro 4 Upstage	Aug 2026	41.6	52.7	—	$0.53	$0.30	$1.20	34	1.26	59.88	89.1%	29.2%	43.6%	70.7%	—	—	—	—	—	—	—	57.3%	—	23.3%
79	MiMo-V2-Pro Xiaomi	Mar 2026	41.4	—	—	$0.00	$0.00	$0.00	0	0.00	0.00	87.0%	30.4%	42.5%	65.7%	68.8%	95.0%	40.9%	—	—	—	—	—	—	—
80	GPT-5.6 Terra (low) OpenAI	Jul 2026	41.3	58.1	—	$4.50	$2.00	$12.00	94	1.50	1.50	84.3%	29.2%	49.2%	68.7%	59.7%	60.5%	43.9%	—	—	—	—	62.5%	—	18.8%
81	Inkling Small Thinking Machines	Jul 2026	41.2	52.9	—	$0.53	$0.30	$1.20	61	2.22	35.00	89.5%	33.3%	48.7%	69.3%	—	—	—	—	—	—	—	55.1%	—	18.8%
82	GPT-5.2 Codex (xhigh) OpenAI	Dec 2025	41.2	—	—	$4.81	$1.75	$14.00	0	0.00	0.00	89.9%	35.7%	54.6%	79.3%	77.6%	92.1%	37.1%	—	—	—	—	—	—	—
83	Qwen3.6 Max Preview Alibaba	Apr 2026	41.1	—	—	$2.93	$1.30	$7.80	0	0.00	0.00	88.8%	30.8%	46.9%	72.0%	76.6%	95.9%	43.9%	—	—	—	—	—	—	—
84	GLM-5.1 (Reasoning) Z AI	Apr 2026	41.0	55.8	—	$2.14	$1.38	$4.40	0	0.00	0.00	86.8%	30.1%	43.8%	68.0%	76.3%	97.7%	43.2%	—	—	—	—	61.8%	—	13.6%
85	GPT-5.4 mini (xhigh) OpenAI	Mar 2026	40.9	56.1	—	$1.69	$0.75	$4.50	0	0.00	0.00	87.5%	28.1%	49.9%	73.0%	73.3%	83.3%	52.3%	—	—	—	—	59.2%	—	25.6%
86	Grok Build 0.1 0616 SpaceXAI	Jun 2026	40.7	51.5	—	$1.25	$1.00	$2.00	0	0.00	0.00	89.5%	38.3%	50.2%	70.0%	—	—	—	—	—	—	—	52.1%	—	13.4%
87	Gemini 3 Pro Preview (high) Google	Nov 2025	40.6	—	95.7	$4.50	$2.00	$12.00	0	0.00	0.00	90.8%	39.7%	56.1%	73.0%	70.4%	87.1%	41.7%	89.8%	91.7%	95.7%	—	—	—	—
88	GLM-5 (Reasoning) Z AI	Feb 2026	40.6	—	—	$1.55	$1.00	$3.20	0	0.00	0.00	82.0%	29.3%	46.2%	70.7%	72.3%	98.2%	43.2%	—	—	—	—	—	—	—
89	Qwen3.6 Plus Alibaba	Apr 2026	40.5	54.5	—	$1.13	$0.50	$3.00	0	0.00	0.00	88.2%	27.8%	40.7%	72.3%	75.2%	97.7%	43.9%	—	—	—	—	61.4%	—	20.8%
90	GPT-5.4 (low) OpenAI	Mar 2026	40.2	—	—	$5.63	$2.50	$15.00	0	0.00	0.00	87.1%	30.8%	50.3%	73.7%	65.9%	74.6%	43.2%	—	—	—	—	—	—	—
91	JT-4.1 Flash 236B A21B China Mobile	Jul 2026	39.9	52.4	—	$0.00	$0.00	$0.00	0	0.00	0.00	84.5%	17.8%	38.2%	70.3%	—	—	—	—	—	—	—	59.6%	—	—
92	Agnes 2.5 Pro Alpha Sapiens AI	Jul 2026	39.7	58.8	—	$0.56	$0.45	$0.90	130	1.44	16.77	87.6%	33.6%	42.2%	73.0%	—	—	—	—	—	—	—	67.0%	—	12.4%
93	GPT-5.4 nano (xhigh) OpenAI	Mar 2026	39.7	56.1	—	$0.46	$0.20	$1.25	0	0.00	0.00	81.7%	28.3%	46.9%	72.0%	75.9%	76.0%	42.4%	—	—	—	—	60.7%	—	27.4%
94	Qwen3.7 Plus Alibaba	Jun 2026	39.4	55.9	—	$0.70	$0.40	$1.60	55	1.48	37.59	90.0%	35.6%	45.5%	69.0%	78.0%	93.0%	47.0%	—	—	—	—	61.0%	—	17.5%
95	GLM-5-Turbo Z AI	Mar 2026	39.1	—	—	$0.00	$0.00	$0.00	0	0.00	0.00	84.7%	27.8%	43.6%	66.7%	73.2%	98.5%	33.3%	—	—	—	—	—	—	—
96	DeepSeek V4 Flash (Reasoning, High Effort) DeepSeek	Apr 2026	39.0	52.0	—	$0.17	$0.13	$0.28	0	0.00	0.00	86.7%	30.3%	42.0%	69.0%	73.5%	95.6%	38.6%	—	—	—	—	56.9%	—	26.2%
97	GPT-5.6 Luna (medium) OpenAI	Jul 2026	38.9	50.7	—	$0.45	$0.20	$1.20	119	2.56	2.56	85.9%	25.8%	45.8%	72.0%	—	—	—	—	—	—	—	53.2%	—	17.7%
98	GPT-5.2 (medium) OpenAI	Dec 2025	38.9	—	96.7	$4.81	$1.75	$14.00	0	0.00	0.00	86.4%	26.7%	46.2%	69.0%	65.2%	74.3%	43.2%	85.9%	89.4%	96.7%	—	—	—	—
99	MiniMax-M2.7 MiniMax	Mar 2026	38.9	52.6	—	$0.53	$0.30	$1.20	0	0.00	0.00	87.4%	29.6%	47.0%	75.3%	75.7%	84.8%	39.4%	—	—	—	—	55.4%	—	9.9%
100	Claude Opus 4.6 (Non-reasoning, High Effort) Anthropic	Feb 2026	38.8	—	—	$10.00	$5.00	$25.00	0	0.00	0.00	84.0%	19.1%	45.7%	62.3%	44.6%	84.8%	48.5%	—	—	—	—	—	—	—
101	Gemini 3 Flash Preview (Reasoning) Google	Dec 2025	38.7	—	97.0	$1.13	$0.50	$3.00	0	0.00	0.00	89.8%	36.6%	50.6%	73.0%	78.0%	80.4%	38.6%	89.0%	90.8%	97.0%	—	—	—	20.8%
102	Nemotron 3 Ultra 550B A55B (Reasoning) NVIDIA	Jun 2026	38.3	49.3	—	$1.14	$0.60	$2.75	70	2.92	35.51	86.7%	28.4%	39.9%	71.0%	81.4%	83.3%	36.4%	—	—	—	—	53.9%	—	14.2%
103	MiMo-V2.5 Xiaomi	Apr 2026	38.0	56.8	—	$0.18	$0.14	$0.28	66	2.15	32.28	84.9%	27.2%	43.1%	68.3%	67.1%	90.6%	41.7%	—	—	—	—	63.7%	—	8.7%
104	Grok 4.20 0309 v2 (Reasoning) SpaceXAI	Apr 2026	38.0	—	—	$1.56	$1.25	$2.50	0	0.00	0.00	91.1%	34.5%	45.6%	62.3%	81.2%	93.0%	37.9%	—	—	—	—	—	—	—
105	Grok 4.3 (high) SpaceXAI	Apr 2026	37.9	42.2	—	$1.56	$1.25	$2.50	0	0.00	0.00	90.1%	37.2%	47.3%	66.0%	81.3%	97.7%	37.9%	—	—	—	—	39.7%	—	12.4%
106	Ling 3.0 Flash InclusionAI	Aug 2026	37.8	50.6	—	$0.11	$0.08	$0.22	371	1.64	7.02	85.5%	23.7%	41.1%	67.0%	—	—	—	—	—	—	—	55.4%	—	27.2%
107	Qwen3.6 27B (Reasoning) Alibaba	Apr 2026	37.7	53.7	—	$1.35	$0.60	$3.60	53	1.44	107.62	84.2%	23.1%	39.8%	73.3%	67.6%	94.2%	34.8%	—	—	—	—	60.7%	—	16.7%
108	GPT-5.1 (high) OpenAI	Nov 2025	37.5	49.4	94.0	$3.44	$1.25	$10.00	0	0.00	0.00	87.3%	28.5%	43.3%	76.7%	72.9%	81.9%	45.5%	87.0%	86.8%	94.0%	—	52.4%	—	15.9%
109	Gemini 3.5 Flash-Lite Google	Jul 2026	37.4	49.3	—	$0.85	$0.30	$2.50	360	6.03	6.03	83.8%	18.8%	40.9%	74.7%	—	—	—	—	—	—	—	53.6%	—	17.5%
110	Solar Open2 250B Upstage	Aug 2026	37.4	44.7	—	$0.00	$0.00	$0.00	0	0.00	0.00	85.7%	28.5%	45.6%	68.3%	—	—	—	—	—	—	—	44.2%	—	21.6%
111	Claude 4.5 Sonnet (Reasoning) Anthropic	Sep 2025	37.4	52.1	88.0	$6.00	$3.00	$15.00	0	0.00	0.00	83.4%	17.8%	44.7%	68.3%	57.3%	78.1%	35.6%	87.5%	71.4%	88.0%	—	55.8%	—	24.5%
112	Grok 4.20 0309 (Reasoning) SpaceXAI	Mar 2026	37.4	—	—	$3.00	$2.00	$6.00	0	0.00	0.00	88.5%	32.4%	44.7%	60.7%	82.9%	96.5%	40.9%	—	—	—	—	—	—	—
113	MiMo-V2-Omni-0327 Xiaomi	Mar 2026	37.3	—	—	$0.00	$0.00	$0.00	0	0.00	0.00	85.5%	22.6%	39.5%	69.7%	67.3%	88.0%	35.6%	—	—	—	—	—	—	—
114	GPT-5 Codex (high) OpenAI	Sep 2025	37.0	—	98.7	$3.44	$1.25	$10.00	0	0.00	0.00	83.7%	27.8%	40.9%	71.0%	74.1%	86.8%	37.9%	86.5%	84.0%	98.7%	—	—	—	—
115	Grok 4.3 (medium) SpaceXAI	Apr 2026	36.9	—	—	$1.56	$1.25	$2.50	118	18.37	18.37	89.0%	30.0%	44.6%	68.3%	83.3%	91.2%	30.3%	—	—	—	—	—	—	—
116	Claude Sonnet 4.6 (Non-reasoning, High Effort) Anthropic	Feb 2026	36.8	—	—	$6.00	$3.00	$15.00	0	0.00	0.00	79.9%	13.3%	46.9%	62.3%	41.2%	79.5%	46.2%	—	—	—	—	—	—	—
117	Grok 4.3 (low) SpaceXAI	Apr 2026	36.3	—	—	$1.56	$1.25	$2.50	110	6.56	6.56	84.3%	18.4%	41.9%	67.7%	81.0%	88.9%	26.5%	—	—	—	—	—	—	—
118	GLM-5.1 (Non-reasoning) Z AI	Apr 2026	36.3	—	—	$2.14	$1.39	$4.40	0	0.00	0.00	83.9%	27.9%	36.1%	53.7%	52.0%	97.1%	35.6%	—	—	—	—	—	—	—
119	Kimi K2.5 (Reasoning) Kimi	Jan 2026	36.0	46.8	—	$1.20	$0.60	$3.00	0	0.00	0.00	87.9%	30.7%	49.0%	73.0%	70.2%	95.9%	34.8%	—	—	—	—	45.7%	—	14.2%
120	MiMo-V2-Omni Xiaomi	Mar 2026	35.9	—	—	$0.00	$0.00	$0.00	0	0.00	0.00	82.8%	22.1%	36.7%	71.7%	53.5%	91.2%	34.8%	—	—	—	—	—	—	—

Independent benchmarks & pricing of LLMs — we publish them here, we don’t produce them. For further analysis and methodology, see artificialanalysis.ai. Updated 1 Sep 2026.

How to read it

Three things that trip people up when they first use this table.

Price is per million tokens

Input and output are priced separately, and output is usually the expensive half. A workflow that reads a lot and writes a little costs far less than the headline number suggests.

Two different kinds of speed

Time to first token is what a person waiting on a chat reply feels. Tokens per second is what matters for a batch job running overnight. Optimising for the wrong one wastes money.

Quality scores are a shortlist, not a verdict

Aggregate indices are a good way to narrow twenty models to three. They are a poor way to choose between those three — for that, test them on your actual task.

Building a workflow? We pick the model for you

We cost and benchmark every automation before it ships — so it runs on the cheapest model that actually holds up on your task.

Browse ready-made workflows Back to the Lab