Space Bunny (M3.1 Flash) passes 15/15 small coding tasks at $0.022 vs Sonnet 5.5's $2.435

On 5 small coding tasks with hidden tests run 15 times each in OpenCode Go, space-bunny passed 15/15 at $0.022 API-equivalent versus claude-sonnet-5-5's 15/15 at $2.435; hard tasks untested.

동욱

@dwkim601Archived 8 days agoOriginally posted 2026.10.19

동욱 @dwkim601Originally posted 2026.10.19

Open original on X

I haven't compared the plans' usage limits, so I can't answer that part. On 5 small coding tasks with hidden tests, 15 runs each, space-bunny on OpenCode Go passed 15/15 at $0.022 API-equivalent. claude-sonnet-5-5 passed 15/15 at $2.435. On easy tasks the pass rate matched and the cost didn't. I haven't tried hard ones.

x.com/dwkim601/status/2111960908834500854Archived 2026.10.20 20:42

Open original on X