live
weekly refresh
basedagi.org
▸ compare

Anthropic: Claude Sonnet 4.5 vs OpenAI: GPT-5.2

Side-by-side benchmark comparison based on independent public data. Scores, pricing, context, and task breakdown.

▸ verdict
Anthropic: Claude Sonnet 4.5
85.7
higher score
vs
OpenAI: GPT-5.2
81.3
Anthropic: Claude Sonnet 4.5 leads with a higher composite benchmark score. It's the stronger choice for general-purpose tasks based on public benchmark data.
▸ score breakdown
TaskAnthropic: Claude Sonnet 4.5OpenAI: GPT-5.2Δ
Overall85.781.3+4.4
Coding59.869.9-10.1
Reasoning45.367.8-22.5
Math
Writing84.379.9+4.4
JSON28.654.2-25.6
▸ specs & pricing
AttributeAnthropic: Claude Sonnet 4.5OpenAI: GPT-5.2
ProviderAnthropicOpenai
Context window1M400K
Input $/M tokens$3.00/M$1.75/M
Output $/M tokens$15.00/M$14.00/M
Weightsproprietaryproprietary
▸ frequently asked

Is Anthropic: Claude Sonnet 4.5 better than OpenAI: GPT-5.2?

Anthropic: Claude Sonnet 4.5 scores higher overall (85.7 vs 81.3) in the benchmark composite. The best choice depends on the specific use case and budget.

Which is cheaper: Anthropic: Claude Sonnet 4.5 or OpenAI: GPT-5.2?

OpenAI: GPT-5.2 is cheaper at $1.75/M input tokens vs $3/M for Anthropic: Claude Sonnet 4.5.

Which model is better for coding?

OpenAI: GPT-5.2 leads on coding with a score of 69.9 vs 59.8 for Anthropic: Claude Sonnet 4.5. This is based on SWE-Bench, Aider Polyglot, and LiveCodeBench data.