All benchmarks
Anthropic · USA · Released 2025-09
Claude Sonnet 4.5
Anthropic's best coding + agentic model, purpose-built for long-horizon computer-use and software engineering tasks.
Access
closed
Context
200,000 tokens
Modalities
text, vision
Languages
95+
Benchmark scores
MMLU
88.5%
Massive Multitask Language Understanding
GPQA
83.4%
Graduate-level science QA (Diamond)
HumanEval
95.4%
Python code completion
SWE-bench
77.2%
Real GitHub issue resolution (Verified)
LiveCodeBench
N/A
Contamination-free coding
AIME
87.0%
American Invitational Mathematics Exam
MMMU
77.8%
Multimodal college-level reasoning
MATH
91.0%
Competition-level math word problems
Scores as reported by the vendor or leading public leaderboards. "N/A" means the score has not been publicly disclosed for this metric.
Explain like I'm 5
Great at writing code and running many steps in a row without getting lost.
Key features
- Extended thinking
- Computer use
- Code editing tools
- Long agent runs (>30h)
Strengths
- Best-in-class SWE-bench
- Superb long-context coding
- Reliable tool use
Limitations
- No native audio
- Closed source
Best for
Autonomous coding agentsRefactoring large codebasesAnalyst workflows
Related models
OpenAI
GPT-5
OpenAI's flagship reasoning + multimodal model with a unified router that switches between fast and deep-thinking modes.
Google DeepMind
Gemini 2.5 Pro
Google's most capable model with native 1M-token context and full multimodal I/O across text, images, audio and video.
xAI
Grok 4
xAI's frontier reasoning model with real-time X data grounding and a heavy multi-agent variant (Grok 4 Heavy).
Meta AI
Llama 4 Maverick
Meta's flagship MoE model — 400B total / 17B active — with native multimodal input and 1M-token context.