Sol is closer to Fable than Opus - I like SlopCodeBench the most as a test - https://github.com/humanlayer/advanced-context-engineering-for-coding-agents/blob/m

Sol is closer to Fable than Opus - I like SlopCodeBench the most as a test - https://github.com/humanlayer/advanced-context-engineering-for-coding-agents/blob/main/benchmarking-sol-fable-kimi-on-slop-code-bench.md https://github.com/humanlayer/advanced-context-engineering-f... You add requirements and make previous tests invisible to see how pigeon brained the model is - Sol and Fable seem to rank the same as Opus tends to fall behind