AI model leaderboards

See top models, labs, apps, and providers on AI Gateway with token, request, and cost trends.

DeepsecBench

DeepsecBench measures how well different AI models find security vulnerabilities in application code, based on the deepsec cyber harness(opens in new tab). The report combines recall and precision into a single benchmark score.

OpenAI

Anthropic

Moonshot AI

DeepSeek

Alibaba

xAI

Google

Z.ai

Thinking Machines

Score vs Cost

Score vs Total Time

Ranked results


1	GPT-5.6 Sol xhighopenai/gpt-5.6-sol	35.58	30.7%	96.3%	71/231	3	$55.98	3h 39m	2.3M	codex
2	Claude Opus 5 maxanthropic/claude-opus-5	32.57	28.1%	88.0%	65/231	10	$127.93	2h 34m	4.5M	claude
3	Claude Opus 5 mediumanthropic/claude-opus-5	28.36	24.7%	70.5%	57/231	26	$31.96	47m 1s	999.7K	claude
4	GPT-5.6 Luna xhighopenai/gpt-5.6-luna	26.79	22.9%	81.2%	53/231	13	$24.59	3h 1m	3.3M	codex
5	GPT-5.6 Sol mediumopenai/gpt-5.6-sol	25.10	21.2%	94.3%	49/231	3	$17.95	30m 52s	703.0K	codex
6	GPT-5.5 xhighopenai/gpt-5.5	21.20	17.7%	95.5%	41/231	2	$43.42	2h 22m	2.2M	codex
7	GPT-5.6 Terra xhighopenai/gpt-5.6-terra	19.21	16.0%	95.3%	37/231	2	$27.81	1h 45m	2.1M	codex
8	Kimi K3 xhighmoonshotai/kimi-k3	17.56	14.7%	77.1%	34/231	11	$12.38	1h 59m	731.9K	pi
9	Claude Sonnet 5 maxanthropic/claude-sonnet-5	17.06	14.3%	76.1%	33/231	11	$45.17	3h 8m	4.6M	claude
10	Deepseek V4 Flash xhighdeepseek/deepseek-v4-flash	16.54	13.9%	74.0%	32/231	13	$5.94	1h 18m	1.5M	pi
11	Qwen3.8 Max xhighalibaba/qwen3.8-max	16.54	13.9%	73.9%	32/231	12	$6.33	1h 52m	342.6K	pi
12	Grok 4.5 mediumxai/grok-4.5	16.54	13.9%	73.3%	32/231	12	$11.04	1h 37m	909.6K	pi
13	Grok 4.6 xhighxai/grok-4.6	16.23	13.4%	100.0%	31/231	0	$22.29	4h 2m	3.5M	pi
14	Qwen3.8 Max mediumalibaba/qwen3.8-max	15.59	13.0%	78.9%	30/231	8	$6.22	2h 3m	383.4K	pi
15	Grok 4.5 xhighxai/grok-4.5	15.58	13.0%	77.8%	30/231	10	$5.60	1h 24m	1.0M	pi
16	Deepseek V4 Flash mediumdeepseek/deepseek-v4-flash	15.48	13.0%	66.7%	30/231	16	$5.06	1h 21m	1.4M	pi
17	GPT-5.5 mediumopenai/gpt-5.5	12.63	10.4%	92.3%	24/231	2	$14.94	24m 5s	751.4K	codex
18	GPT-5.6 Terra mediumopenai/gpt-5.6-terra	12.12	10.0%	92.0%	23/231	2	$5.15	13m 15s	617.7K	codex
19	GPT-5.6 Luna mediumopenai/gpt-5.6-luna	12.08	10.0%	82.8%	23/231	5	$5.23	11m 59s	572.0K	codex
20	Gemini 3.6 Flash mediumgoogle/gemini-3.6-flash	11.56	9.5%	80.0%	22/231	6	$10.74	26m 21s	2.7M	pi
21	Kimi K3 mediummoonshotai/kimi-k3	11.48	9.5%	64.1%	22/231	14	$10.84	1h 29m	676.6K	pi
22	GLM-5.2 mediumzai/glm-5.2	10.46	8.7%	62.5%	20/231	12	$11.09	1h 3m	1.5M	pi
23	GLM-5.2 xhighzai/glm-5.2	9.90	8.2%	53.8%	19/231	18	$51.84	2h 30m	4.7M	pi
24	Gemini 3.6 Flash xhighgoogle/gemini-3.6-flash	8.99	7.4%	77.3%	17/231	5	$5.33	22m 18s	1.8M	pi
25	Claude Opus 4.8 maxanthropic/claude-opus-4.8	6.91	5.6%	81.3%	13/231	3	$7.56	19m 29s	80.8K	claude
26	Claude Opus 4.8 mediumanthropic/claude-opus-4.8	6.88	5.6%	61.9%	13/231	8	$10.81	18m 16s	78.1K	claude
27	Inkling xhighthinkingmachines/inkling	6.32	5.2%	48.0%	12/231	13	$1.93	11m 34s	300.9K	pi
28	Claude Sonnet 5 mediumanthropic/claude-sonnet-5	5.84	4.8%	61.1%	11/231	7	$6.66	15m 47s	442.2K	claude
29	Inkling mediumthinkingmachines/inkling	5.26	4.3%	37.0%	10/231	17	$2.18	10m 31s	509.0K	pi
30	Claude Haiku 4.5 defaultanthropic/claude-haiku-4.5	4.73	3.9%	33.3%	9/231	18	$5.91	29m 40s	912.4K	claude
31	Gemini 3.5 Flash Lite mediumgoogle/gemini-3.5-flash-lite	2.14	1.7%	44.4%	4/231	5	$0.58	1m 46s	459.9K	pi
32	Gemini 3.5 Flash Lite xhighgoogle/gemini-3.5-flash-lite	1.07	0.9%	28.6%	2/231	5	$0.18	1m 7s	406.5K	pi

Agent Stack

Core Platform

Tools

Learn

Build

Explore

AI model leaderboards

DeepsecBench

Score vs Cost

Score vs Total Time

Ranked results