LLM Model Comparison
Compare 44 major language models across pricing, context windows, capabilities, and performance metrics.
Showing 44 of 44 models
44
Total Models
12
Companies
$0.05-$75
Price Range
10M tokens
Max Context
Kimi K2 Thinking
Moonshot AI
Premium
Input:$0.80/M
Output:$3.20/M
Context:256K tokens
Max Output:32K tokens
Latency:Medium
Streaming
Functions
Best for:
Beats GPT-5
71.3% SWE-bench
200-300 tool calls autonomy
Claude Opus 4.1
Anthropic
Premium
Input:$15.00/M
Output:$75.00/M
Context:1M tokens
Max Output:8K tokens
Latency:Medium
Multimodal
Streaming
Functions
Caching
Best for:
Best coding (72.5% SWE-bench)
Long-running tasks
Agent workflows
Gemini 2.5 Pro
Premium
Input:$1.25*/M
Output:$10.00*/M
Cached: $0.31/M
Context:1M tokens
Max Output:8K tokens
Latency:Fast
Multimodal
Streaming
Functions
Caching
Best for:
#1 on LMArena
86.4 GPQA reasoning
Deep Think mode
GPT-5.1 Instant
OpenAI
Premium
Input:$1.25/M
Output:$10.00/M
Cached: $0.125/M
Context:272K tokens
Max Output:128K tokens
Latency:Fast
Multimodal
Streaming
Functions
Caching
Best for:
Adaptive reasoning
Warmer & conversational
More accurate
GPT-5.1 Thinking
OpenAI
Premium
Input:$1.25/M
Output:$10.00/M
Cached: $0.125/M
Context:272K tokens
Max Output:128K tokens
Latency:Adaptive
Multimodal
Streaming
Functions
Caching
Best for:
Advanced reasoning
Faster on simple tasks
More persistent on complex
GPT-5
OpenAI
Premium
Input:$1.25/M
Output:$10.00/M
Cached: $0.125/M
Context:272K tokens
Max Output:128K tokens
Latency:Fast
Multimodal
Streaming
Functions
Caching
Best for:
Software-on-demand
Multimodal
88.4% GPQA
GPT-5 Codex
OpenAI
Premium
Input:$1.25/M
Output:$10.00/M
Cached: $0.125/M
Context:272K tokens
Max Output:128K tokens
Latency:Adaptive
Streaming
Functions
Caching
Best for:
Agentic coding
7+ hour autonomy
Code review
GPT-4o
OpenAI
Premium
Input:$2.50/M
Output:$10.00/M
Context:128K tokens
Max Output:16K tokens
Latency:Fast
Multimodal
Streaming
Functions
Best for:
Flagship general-purpose
Multimodal
Versatile
o3-pro
OpenAI
Premium
Input:$15.00/M
Output:$60.00/M
Context:200K tokens
Max Output:100K tokens
Latency:Slow (reasoning)
Functions
Best for:
Most capable reasoning
Math/science
PhD-level tasks
Grok 4
xAI
Premium
Input:$3.00/M
Output:$15.00/M
Cached: $0.75/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Multimodal
Streaming
Functions
Caching
Best for:
Real-time web search
Trained on Colossus
Multimodal
Mistral Large
Mistral
Premium
Input:$2.00/M
Output:$6.00/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Best for:
European AI leader
123B params
Strong multilingual
Gemini 1.5 Pro
Premium
Input:$1.25/M
Output:$5.00/M
Cached: $0.31/M
Context:2M tokens
Max Output:8K tokens
Latency:Fast
Multimodal
Streaming
Functions
Caching
Best for:
2M context!
Multimodal
Long documents
Claude Sonnet 4.5
Anthropic
Mid-tier
Input:$3.00/M
Output:$15.00/M
Cached: $0.30/M
Context:1M tokens
Max Output:8K tokens
Latency:Fast
Multimodal
Streaming
Functions
Caching
Best for:
Best coding/agents
72.7% SWE-bench
90% cache savings
GPT-4.1
OpenAI
Mid-tier
Input:$2.00/M
Output:$8.00/M
Cached: $0.50/M
Context:1M tokens
Max Output:64K tokens
Latency:Fast
Multimodal
Streaming
Functions
Caching
Best for:
1M context
Long conversations
Improved reasoning
Grok 3
xAI
Mid-tier
Input:$3.00/M
Output:$15.00/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Multimodal
Streaming
Functions
Best for:
Real-time search
Function calling
Fast inference
o4-mini
OpenAI
Mid-tier
Input:$1.10/M
Output:$4.40/M
Context:200K tokens
Max Output:100K tokens
Latency:Medium (reasoning)
Multimodal
Functions
Best for:
Fast reasoning
Best on AIME 2024/2025
Math/coding
Claude Haiku 4.5
Anthropic
Mid-tier
Input:$1.00/M
Output:$5.00/M
Cached: $0.10/M
Context:200K tokens
Max Output:8K tokens
Latency:Very Fast
Multimodal
Streaming
Functions
Caching
Best for:
Fastest
Near Sonnet quality
90% cache savings
Qwen 2.5 72B
Alibaba
Mid-tier
Input:$1.00/M
Output:$1.00/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Best for:
72.7B params
29 languages
Apache 2.0
DeepSeek V3.1
DeepSeek
Cost-effective
Input:$0.56/M
Output:$1.68/M
Cached: $0.07/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Caching
Best for:
Hybrid thinking/non-thinking
671B params
82.6% HumanEval
GPT-5 Mini
OpenAI
Cost-effective
Input:$0.25/M
Output:$2.00/M
Cached: $0.025/M
Context:272K tokens
Max Output:64K tokens
Latency:Fast
Multimodal
Streaming
Functions
Caching
Best for:
GPT-5 quality
272K context
Multimodal
DeepSeek R1
DeepSeek
Cost-effective
Input:$0.55/M
Output:$2.19/M
Cached: $0.14/M
Context:128K tokens
Max Output:32K tokens
Latency:Medium (reasoning)
Streaming
Caching
Best for:
Reasoning model
27x cheaper than o1
MIT license
Qwen3-235B
Alibaba
Cost-effective
Input:$0.35/M
Output:$0.60/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Best for:
235B params (22B active)
Hybrid thinking
Apache 2.0
Qwen3-32B
Alibaba
Cost-effective
Input:$0.20/M
Output:$0.40/M
Context:128K tokens
Max Output:32K tokens
Latency:Very Fast
Streaming
Functions
Best for:
Outperforms o1-mini
Strong reasoning
Apache 2.0
Mistral Medium 3
Mistral
Cost-effective
Input:$0.40/M
Output:$2.00/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Best for:
8x cheaper than competitors
EU hosting
Function calling
Mistral Small 3
Mistral
Cost-effective
Input:$0.20/M
Output:$0.60/M
Context:128K tokens
Max Output:32K tokens
Latency:Very Fast
Streaming
Functions
Best for:
24B params
Fast inference
EU compliant
GPT-OSS-120B
OpenAI
Cost-effective
Input:Self-hosted/M
Output:Self-hosted/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Best for:
Matches o4-mini
117B total (5.1B active)
Open weights
Llama 4 Maverick
Meta
Cost-effective
Input:$0.50/M
Output:$0.77/M
Context:1M tokens
Max Output:32K tokens
Latency:Fast
Multimodal
Streaming
Functions
Best for:
400B params MoE
Multimodal
Open weights
Grok 3 Mini
xAI
Cost-effective
Input:$0.30/M
Output:$0.50/M
Context:128K tokens
Max Output:16K tokens
Latency:Very Fast
Streaming
Functions
Best for:
Fast
Cost-efficient
Function calling
GLM 4.6
Zhipu AI
Cost-effective
Input:$0.30/M
Output:$0.55/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Best for:
355B MoE
Bilingual (CN/EN)
MIT license
Llama 3.3 70B
Meta
Cost-effective
Input:$0.35/M
Output:$0.55/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Best for:
Similar to 3.1 405B
Open weights
Cost-efficient
Gemini 2.0 Flash
Cost-effective
Input:$0.10/M
Output:$0.40/M
Context:1M tokens
Max Output:8K tokens
Latency:Very Fast
Multimodal
Streaming
Functions
Best for:
1M context
Native tool use
Multimodal
Gemini 1.5 Flash
Cost-effective
Input:$0.075/M
Output:$0.30/M
Cached: $0.01875/M
Context:1M tokens
Max Output:8K tokens
Latency:Very Fast
Multimodal
Streaming
Functions
Caching
Best for:
1M context
Fast
Prompt caching
Llama 4 Scout
Meta
Ultra-cheap
Input:$0.11/M
Output:$0.34/M
Context:10M tokens
Max Output:32K tokens
Latency:Fast
Multimodal
Streaming
Functions
Best for:
10M context!
Multimodal
Open weights
GPT-5 Nano
OpenAI
Ultra-cheap
Input:$0.05/M
Output:$0.40/M
Cached: $0.005/M
Context:272K tokens
Max Output:32K tokens
Latency:Very Fast
Multimodal
Streaming
Functions
Caching
Best for:
High throughput
Simple tasks
272K context
Llama 3.1 405B
Meta
Ultra-cheap
Input:$0.30/M
Output:$0.50/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Best for:
Largest open model
Multilingual
128K context
Llama 3.2 90B
Meta
Ultra-cheap
Input:$0.20/M
Output:$0.40/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Multimodal
Streaming
Functions
Best for:
First multimodal Llama
Image+text
Open weights
Llama 3.1 70B
Meta
Ultra-cheap
Input:$0.08/M
Output:$0.15/M
Context:128K tokens
Max Output:32K tokens
Latency:Very Fast
Streaming
Functions
Best for:
Popular for production
Open weights
Strong reasoning
IBM Granite 4.0 Nano 1.5B
IBM
Local/Edge
Input:Self-hosted/M
Output:Self-hosted/M
Context:128K tokens
Max Output:32K tokens
Latency:Fast
Streaming
Functions
Best for:
Runs in browser
70% memory reduction
ISO 42001
IBM Granite 4.0 Nano 350M
IBM
Local/Edge
Input:Self-hosted/M
Output:Self-hosted/M
Context:128K tokens
Max Output:32K tokens
Latency:Very Fast
Streaming
Functions
Best for:
Ultra-compact
Browser-runnable
Enterprise certified
Phi-4
Microsoft
Local/Edge
Input:Self-hosted/M
Output:Self-hosted/M
Context:16K tokens
Max Output:4K tokens
Latency:Very Fast
Streaming
Functions
Best for:
SOTA small model
Reasoning
Runs on laptop
Phi-4-multimodal
Microsoft
Local/Edge
Input:Self-hosted/M
Output:Self-hosted/M
Context:16K tokens
Max Output:4K tokens
Latency:Very Fast
Multimodal
Streaming
Functions
Best for:
Text/image/audio/video
5.6B params
20+ languages
Florence-2
Microsoft
Local/Edge
Input:Self-hosted/M
Output:Self-hosted/M
Context:N/A
Max Output:N/A
Latency:Fast
Multimodal
Best for:
Vision foundation model
OCR
Object detection
Llama 3.2 3B
Meta
Local/Edge
Input:Self-hosted/M
Output:Self-hosted/M
Context:128K tokens
Max Output:32K tokens
Latency:Very Fast
Streaming
Functions
Best for:
Compact
Mobile-friendly
Open weights
Llama 3.2 1B
Meta
Local/Edge
Input:Self-hosted/M
Output:Self-hosted/M
Context:128K tokens
Max Output:32K tokens
Latency:Very Fast
Streaming
Best for:
Ultra-compact
Edge deployment
Open weights
Need Help Choosing?
Check out our detailed comparison table or explore individual model pages for in-depth specifications.