technical reference
AI Models
23 modelos ranqueados por benchmark — filtre por categoria de uso.
Claude Opus 4.6 Anthropic · opus
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
Claude Sonnet 4.6 Anthropic · sonnet
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
Claude Opus 4.5 Anthropic · opus
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
GPT-5.4 Pro OpenAI · pro
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
GPT-5.4 OpenAI
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
GPT-5.3 Codex OpenAI · codex
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
GPT-5.3 Chat OpenAI · chat
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
GPT-5.2 Pro OpenAI · pro
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
GPT-5.2 OpenAI
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
o3 OpenAI
1424 Arena ELO
96.7% MATH
96.7% SimpleQA
Price N/A
Latency N/A
o4-mini OpenAI · mini
1391 Arena ELO
97.5% MATH
97.3% HumanEval
Price N/A
Latency N/A
Gemini 3.1 Pro Google · pro
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
Gemini 3.1 Flash Lite Google · flash-lite
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
Gemini 3 Flash Google · flash
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
Grok 3 xAI
1423 Arena ELO
93.3% MATH
92.7% MMLU
Price N/A
Latency N/A
Grok 3 Mini xAI · mini
1366 Arena ELO
95.8% MATH
80.4% LiveCodeBench
Price N/A
Latency N/A
DeepSeek V3.2 DeepSeek
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
DeepSeek V3.2 Speciale DeepSeek · speciale
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
DeepSeek R1 DeepSeek
1358 Arena ELO
97.3% MATH
92.0% HumanEval
Price N/A
Latency N/A
Kimi K2.5 Moonshot AI
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
MiniMax M2.5 MiniMax
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
MiniMax M2.1 MiniMax
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A
MiniMax M2-her MiniMax · her
— SWE-bench
— HumanEval
— MMLU
Price N/A
Latency N/A