LLM Leaderboard - Comparison of over 100 AI models from OpenAI, Google, DeepSeek & others
Comparison and ranking the performance of over 100 AI models (LLMs) across key metrics including intelligence, 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.
HIGHLIGHTS
Claude Opus 4.6 (Adaptive) and 
GPT-5.2 (xhigh) are the highest intelligence models, followed by Claude Opus 4.5 & 
GLM-5.Output Speed (tokens/s):
Gemini 2.5 Flash-Lite (Sep) (506 t/s) and 
Granite 3.3 8B (495 t/s) are the fastest models, followed by Gemini 2.5 Flash-Lite (Sep) & 
Nova Micro.Latency (seconds):
Apriel-v1.5-15B-Thinker (0.17s) and 
DeepSeek-OCR (0.19s) are the lowest latency models, followed by Apriel-v1.6-15B-Thinker & 
NVIDIA Nemotron Nano 12B v2 VL.Price ($ per M tokens): Gemma 3n E4B ($0.03) and DeepSeek-OCR ($0.05) are the cheapest models, followed by 
Llama 3.2 1B & Nova Micro.Context Window: Llama 4 Scout (10m) and 
Grok 4.1 Fast (2m) are the largest context window models, followed by Grok 4.1 Fast & Gemini 2.0 Pro Experimental.Key definitions
Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).
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).
Time to first token received, in seconds, after API request sent. For reasoning models which share reasoning tokens, this will be the first reasoning token. For models which do not support streaming, this represents time to receive the completion.
Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).
Price per token generated by the model (received from the API), represented as USD per million Tokens.
Price per token included in the request/message sent to the API, represented as USD per million Tokens.
Metrics are 'live' and are based on the past 72 hours of measurements, measurements are taken 8 times a day for single requests and 2 times per day for parallel requests.
