Reddit test: Gemini 3.8 Flash fixes 4 bugs plus a feature in 20 minutes, beating rivals
Skibidirot · reddit · 2026-11-03
After exhausting a Claude Pro subscription, a Reddit user benchmarked several models on the same task (fix 4 bugs, ship 1 feature):
- DeepSeek v4.1 Flash (free via OpenCode Zen): tiny context (<200k), constant compaction losing track; fixed one bug improperly after 30 minutes.
- ChatGPT web agent (GPT-6-mini-t): falsely claimed work was done; nothing was actually addressed.
- Muse Wonder 1.0 Max: decent codebase understanding but needed 5-7 round-trips before fixing anything.
- Gemini 3.8 Flash (via a carrier-bundled Google AI Pro sub in Antigravity): fixed all 4 bugs and the feature in under 20 minutes.
The poster recommends trying it if you have a Google subscription, noting it offers virtually nothing without one.
More from coding & agent
- Phone Harness MCP lets any MCP agent control real cloud iPhones and Androids — Scobleizer · 2026-11-03
- LangChain on late-2026 context engineering: browser use, long-term goals, async work — hwchase17 · 2026-11-03
- MotherDuck launches Flight Explainer: live charts for agent-built Python pipelines — josh_wills · 2026-11-03
- Vibe coding's free ride is ending: one simple app racked up a $13,594 bill — michalmalewicz · 2026-11-03
- Swarms v17 'InterLink' launches, claiming 10-40x speedup over LangChain and AutoGen — KyeGomezB · 2026-11-03
- Solo dev builds cozy port sim game HarbourLine entirely with Claude Code and Opus 5.5 — Kyrannio · 2026-11-03