Hands-on with Grok-4.7-high: distributed lock test deadlocks, loses to GPT-5.6-sol-medium on real projects
karminski3 · x · 2026-09-23
After a full day with Grok-4.7-high, the verdict: it loses to GPT-5.6-sol-medium on engineering work — "wrote lots of garbage code."
- One-shot tasks are fine, but real projects fall apart;
- Key weaknesses: narrow context, poor code intuition and experience, frequently missing edge cases;
- A quick repro: ask it to write a distributed lock — it deadlocks within minutes and can't fix itself.
More from coding & agent
- Vercel CEO Backs px0, a Lightweight IDE Built for Reviewing Agent-Written Code — arpit_bhayani · 2026-09-23
- Matt Pocock: Stop chasing model releases, improve your agent's harness instead — mattpocockuk · 2026-09-23
- Uncle Bob: AI changes nothing—complexity, not tooling, still makes software slow — blaizedsouza · 2026-09-23
- GBrain: plug your own memory, tools, and skills into any AI — garrytan · 2026-09-23
- Podcast: building a playable game with $8 of parts and AI assistance — aishashok14 · 2026-09-23
- Agent kept searching but never opened the source: four runs expose a hidden failure mode — memokris · 2026-09-23