LLM Leaderboard - Comparison of GPT-4o, Llama 3, Mistral, Gemini and over 30 models

Comparison and ranking the performance of over 30 AI models (LLMs) across key metrics including quality, price, performance and speed (output speed - tokens per second & latency - TTFT), context window & others. For more details including relating to our methodology, see our FAQs.

For comparison of API Providers hosting the models see

HIGHLIGHTS

Quality:

o1-preview and o1-mini logo

o1-mini are the highest quality models, followed by Gemini 1.5 Pro (Sep) logo

Gemini 1.5 Pro (Sep) & Claude 3.5 Sonnet (Oct) logo

Claude 3.5 Sonnet (Oct).Output Speed (tokens/s): Llama 3.2 1B logo

Llama 3.2 1B (552 t/s) and Gemini 1.5 Flash (May) logo

Gemini 1.5 Flash (May) (311 t/s) are the fastest models, followed by Gemini 1.5 Flash-8B logo

Gemini 1.5 Flash-8B & o1-mini logo

o1-mini.Latency (seconds): Aya Expanse 32B logo

Aya Expanse 32B (0.22s) and Aya Expanse 8B logo

Aya Expanse 8B (0.29s) are the lowest latency models, followed by Llama 3.2 11B (Vision) logo

Llama 3.2 11B (Vision) & Gemini 1.5 Flash (May) logo

Gemini 1.5 Flash (May).Price ($ per M tokens): Gemini Experimental (Nov) logo

Gemini Experimental (Nov) ($0.00) and Ministral 3B logo

Ministral 3B ($0.04) are the cheapest models, followed by Llama 3.2 1B logo

Llama 3.2 1B & OpenChat 3.5 logo

OpenChat 3.5.Context Window: Gemini 1.5 Pro (Sep) logo

Gemini 1.5 Pro (Sep) (2m) and Gemini 1.5 Pro (May) logo

Gemini 1.5 Pro (May) (2m) are the largest context window models, followed by Gemini 1.5 Flash-8B logo

Gemini 1.5 Flash-8B & Gemini 1.5 Flash (Sep) logo

Gemini 1.5 Flash (Sep).

		Quality	Price	Output tokens/s	Latency
						Further Analysis
o1-preview	128k	86	$26.25	142.7	25.24
o1-mini	128k	84	$5.25	227.9	9.64
GPT-4o (Aug '24)	128k	78	$4.38	82.4	0.66
GPT-4o (May '24)	128k	78	$7.50	97.1	0.64
GPT-4o (Nov '24)	128k	73	$4.38	112.7	0.38
GPT-4o mini	128k	73	$0.26	114.8	0.68
Llama 3.3 70B	128k	74	$0.75	171.0	0.38
Llama 3.1 405B	128k	74	$3.50	28.9	0.72
Llama 3.1 70B	128k	68	$0.75	72.6	0.43
Llama 3.2 90B (Vision)	128k	68	$0.81	36.4	0.42
Llama 3.2 11B (Vision)	128k	54	$0.18	123.6	0.29
Llama 3.1 8B	128k	54	$0.10	172.6	0.34
Llama 3.2 3B	128k	49	$0.08	199.4	0.32
Llama 3.2 1B	128k	26	$0.05	552.5	0.33
Gemini 1.5 Pro (Sep)	2m	81	$2.19	59.3	0.84
Gemini 1.5 Flash (Sep)	1m	72	$0.13	188.3	0.35
Gemma 2 27B	8k	61	$0.26	48.6	0.59
Gemma 2 9B	8k	55	$0.12	165.5	0.43
Gemini Experimental (Nov)	33k		$0.00	40.4	2.62
Gemini 1.5 Pro (May)	2m		$2.19	66.0	0.74
Gemini 1.5 Flash (May)	1m		$0.13	310.7	0.30
Gemini 1.5 Flash-8B	1m		$0.07	282.4	0.36
Claude 3.5 Sonnet (Oct)	200k	80	$6.00	55.5	0.95
Claude 3.5 Sonnet (June)	200k	76	$6.00	54.5	0.91
Claude 3 Opus	200k	70	$30.00	26.5	1.78
Claude 3.5 Haiku	200k	68	$1.60	65.1	0.79
Claude 3 Haiku	200k	55	$0.50	122.5	0.49
Pixtral Large	128k	74	$3.00	37.2	0.81
Mistral Large 2 (Jul '24)	128k	74	$3.00	34.1	0.50
Mistral Large 2 (Nov '24)	128k	74	$3.00	45.0	0.69
Mistral Small (Sep '24)	128k	61	$0.30	63.3	0.59
Mixtral 8x22B	65k	61	$1.20	86.0	0.62
Pixtral 12B	128k	56	$0.13	70.7	0.50
Ministral 8B	128k	56	$0.10	132.3	0.54
Mistral NeMo	128k	54	$0.09	115.3	0.54
Ministral 3B	128k	53	$0.04	165.4	0.57
Mixtral 8x7B	33k	41	$0.50	123.7	0.37
Codestral-Mamba	256k	33	$0.25	96.0	0.69
Command-R+	128k	55	$5.19	50.1	0.48
Command-R+ (Apr '24)	128k	45	$6.00	48.0	0.53
Command-R (Mar '24)	128k	36	$0.75	107.6	0.36
Command-R	128k		$0.51	111.7	0.33
Aya Expanse 32B	128k		$0.75	120.9	0.22
Aya Expanse 8B	8k		$0.75	143.9	0.29
Sonar 3.1 Large	131k		$1.00	54.2	0.33
Sonar 3.1 Small	131k		$0.20	147.5	0.33
Grok Beta	8k	72	$7.50	61.5	0.46
Nova Pro	300k	75	$1.40	92.7	0.42
Nova Lite	300k	70	$0.10	147.7	0.38
Nova Micro	130k	66	$0.06	197.0	0.36
Phi-3 Medium 14B	128k		$0.30	50.1	0.43
Solar Mini	4k	47	$0.15	85.4	1.10
DBRX	33k	46	$1.16	82.7	0.47
Llama 3.1 Nemotron 70B	128k	72	$0.27	47.3	0.60
Reka Flash	128k	59	$0.35	32.8	1.27
Reka Core	128k	58	$2.00	15.0	1.16
Reka Flash (Feb '24)	128k	46	$0.35	31.2	0.92
Reka Edge	64k	31	$0.10	35.1	0.92
Jamba 1.5 Large	256k		$3.50	51.1	0.70
Jamba 1.5 Mini	256k		$0.25	82.7	0.49
DeepSeek-Coder-V2	128k	71	$0.17	17.0	0.95
DeepSeek-V2	128k		$0.17	17.1	0.91
DeepSeek-V2.5	128k		$1.09	14.7	0.88
Qwen2.5 72B	131k	77	$0.40	56.4	0.56
Qwen2.5 Coder 32B	131k	72	$0.80	58.2	0.37
Qwen2 72B	128k	72	$0.63	46.0	0.38
Yi-Large	32k	61	$3.00	66.5	0.43
GPT-4 Turbo	128k	75	$15.00	39.5	1.23
GPT-3.5 Turbo	16k	53	$0.75	121.5	0.61
GPT-4	8k		$37.50	26.1	0.70
Llama 3 70B	8k	47	$0.89	47.1	0.45
Llama 3 8B	8k	45	$0.15	116.5	0.33
Llama 2 Chat 13B	4k		$0.56	41.7	0.45
Llama 2 Chat 7B	4k		$0.33	124.4	0.49
Gemini 1.0 Pro	33k		$0.75	103.0	1.29
Claude 3 Sonnet	200k	57	$6.00	63.0	0.75
Mistral Small (Feb '24)	33k	59	$1.50	54.0	0.42
Mistral Large (Feb '24)	33k	56	$6.00	38.0	0.50
Mistral 7B	33k	28	$0.18	99.3	0.34
Codestral	33k		$0.30	82.0	0.56
Mistral Medium	33k		$4.09	43.9	0.63
OpenChat 3.5	8k	44	$0.06	77.3	0.34
Jamba Instruct	256k		$0.55	75.4	0.53

Key definitions

Back to Navigation

Artificial Analysis Quality Index: Average result across our evaluations covering different dimensions of model intelligence. Currently includes MMLU, GPQA, Math & HumanEval. OpenAI o1 model figures are preliminary and are based on figures stated by OpenAI. See methodology for more details.

Context window: Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.

Price: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).

Output price: Price per token generated by the model (received from the API), represented as USD per million Tokens.

Input price: Price per token included in the request/message sent to the API, represented as USD per million Tokens.

Time period: Metrics are 'live' and are based on the past 14 days of measurements, measurements are taken 8 times a day for single requests and 2 times per day for parallel requests.

Models compared: OpenAI: GPT 4o Audio, GPT 4o Realtime, GPT 4o Speech Pipeline, GPT-3.5 Turbo, GPT-3.5 Turbo (0125), GPT-3.5 Turbo (1106), GPT-3.5 Turbo Instruct, GPT-4, GPT-4 Turbo, GPT-4 Turbo (0125), GPT-4 Vision, GPT-4o (Aug '24), GPT-4o (May '24), GPT-4o (Nov '24), GPT-4o mini, o1-mini, and o1-preview, Meta: Code Llama 70B, Llama 2 Chat 13B, Llama 2 Chat 70B, Llama 2 Chat 7B, Llama 3 70B, Llama 3 8B, Llama 3.1 405B, Llama 3.1 70B, Llama 3.1 8B, Llama 3.2 11B (Vision), Llama 3.2 1B, Llama 3.2 3B, Llama 3.2 90B (Vision), and Llama 3.3 70B, Google: Gemini 1.0 Pro, Gemini 1.5 Flash (May), Gemini 1.5 Flash (Sep), Gemini 1.5 Flash-8B, Gemini 1.5 Pro (May), Gemini 1.5 Pro (Sep), Gemini 2.0 Flash (Experimental), Gemini Experimental (Nov), Gemma 2 27B, Gemma 2 9B, and Gemma 7B, Anthropic: Claude 2.0, Claude 2.1, Claude 3 Haiku, Claude 3 Opus, Claude 3 Sonnet, Claude 3.5 Haiku, Claude 3.5 Sonnet (June), Claude 3.5 Sonnet (Oct), and Claude Instant, Mistral: Codestral, Codestral-Mamba, Ministral 3B, Ministral 8B, Mistral 7B, Mistral Large (Feb '24), Mistral Large 2 (Jul '24), Mistral Large 2 (Nov '24), Mistral Medium, Mistral NeMo, Mistral Small (Feb '24), Mistral Small (Sep '24), Mixtral 8x22B, Mixtral 8x7B, Pixtral 12B, and Pixtral Large, Cohere: Aya Expanse 32B, Aya Expanse 8B, Command, Command Light, Command-R, Command-R (Mar '24), Command-R+ (Apr '24), and Command-R+, Perplexity: PPLX-70B Online, PPLX-7B-Online, Sonar 3.1 Huge, Sonar 3.1 Large, Sonar 3.1 Small , Sonar Large, and Sonar Small, xAI: Grok Beta and Grok-1, OpenChat: OpenChat 3.5, Amazon: Nova Lite, Nova Micro, and Nova Pro, Microsoft Azure: Phi-3 Medium 14B and Phi-3 Mini, Upstage: Solar Mini, Solar Pro, and Solar Pro (Nov '24), Databricks: DBRX, NVIDIA: Llama 3.1 Nemotron 70B, IBM: Granite 3.0 2B, OpenVoice: Granite 3.0 8B, Reka AI: Reka Core, Reka Edge, Reka Flash (Feb '24), and Reka Flash, Other: LLaVA-v1.5-7B, AI21 Labs: Jamba 1.5 Large, Jamba 1.5 Mini, and Jamba Instruct, DeepSeek: DeepSeek-Coder-V2, DeepSeek-V2, and DeepSeek-V2.5, Snowflake: Arctic, Alibaba: QwQ 32B-Preview, Qwen2 72B, Qwen2.5 Coder 32B, and Qwen2.5 72B, and 01.AI: Yi-Large.