Top 25 AI Models — Complete Leaderboard
Compare the top 25 AI language models ranked by coding, reasoning, and tool-use benchmarks. All pricing and specifications verified against official provider documentation.
Need the cheapest strong model?
Start with DeepSeek V4 Pro, MiniMax M2.5, or Gemini 3.1 Flash-Lite if cost matters more than bragging rights.
See quick picks →Need self-hosted privacy?
2 models in this leaderboard support open-weights / local-first workflows.
Explore local models →Want the full tradeoff?
Compare open weights vs proprietary models before you commit to an API provider.
Read the guide → 1
GPT-5.5
OpenAI • 1M context
9.8
Coding
9.8
Reasoning
9.7
Tool Use
Complex reasoningCodingProfessional workflows
$5
Input $/M tokens
$30
Output $/M tokens
9.78
Weighted Score
2
Claude Opus 4.8
Anthropic • 1M context
9.8
Coding
9.8
Reasoning
9.6
Tool Use
Complex reasoningAgentic codingCritical decisions
$5
Input $/M tokens
$25
Output $/M tokens
9.75
Weighted Score
3
GPT-5.4
OpenAI • 1M context
9.8
Coding
9.5
Reasoning
9.7
Tool Use
CodingAgentsTool integration
$2.5
Input $/M tokens
$15
Output $/M tokens
9.67
Weighted Score
4
Gemini 3.1 Pro Preview
Google • 1M context
9.5
Coding
9.5
Reasoning
9.3
Tool Use
Multimodal tasksLong contextSearch integration
$2
Input $/M tokens
$12
Output $/M tokens
9.45
Weighted Score
5
Gemini 3.5 Flash
Google • 1M context
9.4
Coding
9.4
Reasoning
9.4
Tool Use
Fast multimodal agentsSearch groundingHigh-volume workflows
$1.5
Input $/M tokens
$9
Output $/M tokens
9.40
Weighted Score
6
Claude Sonnet 4.6
Anthropic • 1M context
9.4
Coding
9.3
Reasoning
9.1
Tool Use
Balanced performanceProduction workloadsCost-efficient
$3
Input $/M tokens
$15
Output $/M tokens
9.29
Weighted Score
7
GPT-5.2-Codex
OpenAI • 400K context
9.7
Coding
9.3
Reasoning
9.4
Tool Use
Coding-focused tasksType inferenceAgentic coding
$1.75
Input $/M tokens
$14
Output $/M tokens
9.48
Weighted Score
8
GLM-5
Zhipu AI • 205K context
9.2
Coding
9.3
Reasoning
9
Tool Use
Bilingual (CN/EN)Value-focusedEnterprise
$0.5
Input $/M tokens
$2
Output $/M tokens
9.18
Weighted Score
9
Llama 4 Maverick
Meta • 1M context
9
Coding
9.1
Reasoning
8.7
Tool Use
Open weightsMultimodalCustomizable
$2
Input $/M tokens
$8
Output $/M tokens
8.96
Weighted Score
10
DeepSeek V4 Pro
DeepSeek • 1M context
9.1
Coding
9.2
Reasoning
8.9
Tool Use
Budget codingHigh-volumeCost-sensitive
$0.435
Input $/M tokens
$0.87
Output $/M tokens
9.08
Weighted Score
11
GPT-5.2
OpenAI • 400K context
9.3
Coding
9.2
Reasoning
9
Tool Use
General-purposeBalanced tasks
$1.75
Input $/M tokens
$14
Output $/M tokens
9.19
Weighted Score
12
Mistral Medium 3.5
Mistral • 128K context
9.1
Coding
9.1
Reasoning
8.9
Tool Use
European complianceAgentic codingEnterprise
$2
Input $/M tokens
$6
Output $/M tokens
9.05
Weighted Score
13
Kimi K2.5
Moonshot AI • 256K context
9.4
Coding
9.3
Reasoning
9.2
Tool Use
Visual codingLong contextAgent workflows
$0.6
Input $/M tokens
$2.5
Output $/M tokens
9.32
Weighted Score
14
MiniMax M2.5
MiniMax • 196K context
9.1
Coding
9.2
Reasoning
8.9
Tool Use
Real-world productivityCost-sensitiveHigh-volume
$0.3
Input $/M tokens
$1.2
Output $/M tokens
9.08
Weighted Score
15
Grok 4.3
xAI • 1M context
9.2
Coding
9.3
Reasoning
9
Tool Use
Long contextWeb searchX platform data
$1.25
Input $/M tokens
$2.5
Output $/M tokens
9.18
Weighted Score
16
Qwen3 Max 2026-01-23
Alibaba • 262K context
9.2
Coding
9.2
Reasoning
8.9
Tool Use
MultilingualEnterpriseChinese language
$1.2
Input $/M tokens
$6
Output $/M tokens
9.13
Weighted Score
17
GPT-OSS-120B
OpenAI • 128K context
9.3
Coding
9.2
Reasoning
9
Tool Use
Self-hostedPrivacyCustomization
$0
Input $/M tokens
$0
Output $/M tokens
9.19
Weighted Score