← tools

Model Comparison

Filter, sort, and compare AI models side by side.

Model ↑ProviderContextInput /1MOutput /1MCodingReasoningSpeedStrengths
Claude Fable 5◉Anthropic1M$10$50
98
99
45
Frontier reasoning, 1M context, adaptive effort
Claude Haiku 4.5Anthropic200K$1$5
78
75
95
Speed, cost efficiency, high volume
Claude Opus 4.6Anthropic200K$15$75
97
98
40
Most capable Claude, complex reasoning, agentic tasks
Claude Opus 4.7◉Anthropic1M$5$25
97
97
50
Deep reasoning, 1M context, agentic coding
Claude Sonnet 4.5Anthropic200K$3$15
92
93
70
Near-Opus intelligence at Sonnet pricing
DeepSeek R1OSSDeepSeek64K$0.56$1.68
85
90
35
Open-source reasoning, competes with o1
DeepSeek V3OSSDeepSeek163K$0.56$1.68
86
82
70
Cheap, strong general-purpose, open weights
Gemini 2.0 FlashGoogle1M$0.1$0.4
78
75
92
Rock-bottom pricing, speed, multimodal
Gemini 2.5 FlashGoogle1.048M$0.3$2.5
82
82
88
Cheap thinking model, massive context
Gemini 2.5 ProGoogle1.048M$1.25$10
90
90
55
Thinking model, coding, huge context
Gemini 3.5 Flash◉Google1.048M$1.5$9
88
89
95
Streaming latency, native audio and video, 1M context
GLM 5.1OSS◉Z.ai202K$0.98$3.08
75
72
60
Long-horizon autonomous coding tasks
GLM-5OSS◉Z.ai202KFree*Free*
87
85
70
Open weights, strong reasoning per dollar, self-hosting
GPT-4.1OpenAI1.047M$2$8
88
86
72
Million-token context, coding, instruction following
GPT-4oOpenAI128K$2.5$10
85
85
75
Multimodal, general purpose
GPT-5OpenAI400K$1.25$10
93
94
55
Massive context, strong all-rounder
GPT-5 MiniOpenAI400K$0.25$2
82
80
80
Budget GPT-5 class, large context
GPT-5 NanoOpenAI400K$0.05$0.4
68
65
96
Cheapest frontier model, on-device capable
GPT-5.4◉OpenAI1.05M$2.5$15
92
90
70
Agentic coding, computer use, long context
GPT-5.5◉OpenAI1.05M$5$30
96
96
60
Strong all-rounder, instant routing, broad tooling
GPT-5.5 Pro◉OpenAI1.05M$30$180
95
96
35
Deepest reasoning, hardest problems
Grok 4.3◉xAI1M$1.25$2.5
91
92
80
Speed at frontier level, low price, 1M context
Kimi k2Moonshot AI131K$0.57$2.3
76
82
50
Long-context reasoning, competitive pricing
Kimi K2.6◉Moonshot AI262K$0.55$3.2
89
87
65
Agentic tool use, long context, price
Laguna XS.2OSS◉Poolside262K$0.1$0.2
82
70
88
Local agentic coding at small footprint
Llama 4 MaverickOSSMeta1.048MFree*Free*
85
83
55
Strong open-source, 1M context, multimodal
Llama 4 ScoutOSSMeta327KFree*Free*
82
80
45
10M context, MoE architecture, self-hostable
MiniMax M2OSS◉Minimax204K$0.26$1
78
70
80
Agentic tool calling and coding at low cost
MiniMax M2.7OSS◉Minimax204K$0.25$1
80
73
80
Agentic coding and reasoning at low cost, open weights
Mistral Large 3Mistral262K$0.5$1.5
83
80
60
Multilingual, code, reasoning, larger context
Mistral Small 3.1Mistral128K$0.35$0.55
72
68
90
Very cheap, strong for its size
Mistral Small 4OSS◉Mistral262K$0.15$0.6
78
76
82
Multimodal + reasoning + coding in one model
Nemotron 3 SuperOSS◉Nvidia1M$0.09$0.45
80
82
88
Efficient MoE agentic reasoning, high throughput
o3OpenAI200K$2$8
90
96
25
Reasoning, maths, science, coding
o3-miniOpenAI200K$1.1$4.4
82
88
55
Budget reasoning model, structured thinking
Qwen 2.5 72BOSSAlibaba32KFree*Free*
78
75
60
Strong multilingual, coding, open weights
Qwen3.6 27BOSS◉Alibaba262K$0.29$3.17
72
70
70
Agentic coding and repo-level comprehension
Qwen3.6 35B A3BOSS◉Alibaba262K$0.14$1
70
69
85
Low-cost fast MoE with multimodal input
Trinity Large ThinkingOSS◉Arcee-Ai262K$0.25$0.8
83
88
75
Frontier open reasoning, long-horizon agents

39 of 39 models shown. Prices per 1M tokens (USD). *Open source models are free to self-host.

scores: editorial, calibrated against public benchmarks · ◉ = seen on the Radar