Rankings & evidence
Compare price, reliability, speed, coverage and compatibility using consistent criteria and verifiable evidence.
Evidence needed to compare price
Match model version, currency, input / output / cache, billing unit and observation date.
Price · No comparable samples yet
Not rankedThere are no qualifying comparable samples in this scope. Browse the directory and choose entries and comparison conditions. Rankings require the relevant rules, time window and sample size.
- OpenAI API →Official API · Check the service details for evidence
- Claude API（Anthropic） →Official API · Check the service details for evidence
- Gemini API（Google AI Studio） →Official API · Check the service details for evidence
- OpenRouter →Aggregator · Check the service details for evidence
- Amazon Bedrock →Cloud platform · Check the service details for evidence
- Microsoft Foundry →Cloud platform · Check the service details for evidence
- SiliconFlow →Aggregator · Check the service details for evidence
- DeepSeek API →Official API · Check the service details for evidence
- Kimi API →Official API · Check the service details for evidence
- Qwen API →Official API · Check the service details for evidence
- GLM API (Zhipu) →Official API · Check the service details for evidence
Third-party rankings & benchmarks
Check the original methodology and current results. External scores are not converted into BuYaoAI rankings.
Compare code-editing runs across model developers, with reasoning settings, tool versions, edit formats and samples recorded.
Artificial Analysis →Compare intelligence scores, text token prices and model release dates.
DeepSeek Research →Explore the original R1 distillation results for math, knowledge and coding. These developer-reported results apply to the stated model versions.
Arena (formerly LMArena) →Explore model comparisons on the official leaderboards. The former domain redirects to arena.ai; results are specific to each task board.
EuroEval →Compare generative tasks across 32 European languages, with mean rank scores and 95% confidence intervals separated by split and prompting method.
LiveBench →Explore results and run costs for long tasks, reasoning, coding, browsing, data analysis, research and document design.
BFCL · Berkeley Function Calling →Compare function calling, web search, memory and multi-turn tasks, alongside run costs and latency.
ARC Prize →Compare abstract reasoning, interactive learning and run costs, separated by benchmark version and framework.
OpenCompass →Compare overall, knowledge, reasoning, math and coding results by quarter and task.
LiveCodeBench →Compare code-generation Pass@1 by problem release date and difficulty, with configurations and evaluation scope retained.