LLM Leaderboard

Ranked models

Compare quality, speed, latency, and cost across available models.

Synced Aug 6, 4:02 PM

Ranked for

	#	Model	Overall Score↓	Quality (80%)		Performance (20%)
	#	Model	Overall Score↓	Arena ELO40% weight	Artificial Analysis Intelligence Index40% weight	Speed10% weight	Latency10% weight
	1	Claude Opus 5 Anthropic	0.898	1,492	60.7	94.0 t/s	2.40s
	2	Kimi K3 Moonshot	0.857	1,485	57.1	98.0 t/s	2.06s
	3	Qwen3.8 Max Alibaba	0.856	1,496	56.2	37.0 t/s	3.17s
	4	GPT-5.6 Sol OpenAI	0.840	1,483	58.9	72.0 t/s	4.97s
	5	GPT-5.5 OpenAI	0.820	1,482	54.8	97.5 t/s	4.09s
	6	Gemini 3.6 Flash Google	0.801	1,483	50.1	92.0 t/s	2.00s
	7	Grok 4.5 SpaceXAI	0.787	1,469	53.8	54.0 t/s	1.02s
	8	GPT-5.6 Luna OpenAI	0.784	—	51.2	105.0 t/s	0.95s
	9	GPT-5.6 Terra OpenAI	0.779	1,468	55.0	94.5 t/s	4.90s
	10	Claude Sonnet 5 Anthropic	0.764	1,462	53.4	95.0 t/s	3.08s
	11	Gemini 3.1 Pro Preview Google	0.757	1,485	46.5	72.0 t/s	4.91s
	12	Muse Spark 1.1 Meta	0.729	1,490	50.6	78.0 t/s	16.28s
	13	MiMo-V2.5-Pro Xiaomi	0.707	1,466	42.2	84.0 t/s	0.49s
	14	DeepSeek V4 Pro DeepSeek	0.705	1,458	44.3	97.0 t/s	0.45s
	15	GLM 5.2 Z.ai	0.687	1,469	39.5	98.0 t/s	2.45s

About these metrics

Score

Composite score (0–1) blending the selected category’s quality metrics with speed and latency. Higher is better, relative to the models currently listed.

Quality (80%)

Arena ELO: Chatbot Arena ELO overall rating.40% weight
Artificial Analysis Intelligence Index: Measures general reasoning and knowledge capabilities.40% weight

Performance (20%)

Speed: Output throughput in tokens per second from the fastest available provider.10% weight
Latency: Time to first token in seconds — lower is better.10% weight

Use the + in the table header to view additional reference benchmarks and pricing metrics.

0 / 6 selected