Power user accuses Gemini coding models of deliberate degradation: 10x tokens, basic mistakes
RossPeili · reddit · 2026-11-01
A Reddit user with three years of daily AI IDE experience alleges frontier labs are deliberately degrading coding models. He says Gemini 2.5 Flash and 3.x once solved repo tasks in one shot while following strict contributing guidelines, but the new 3.8 Flash High burns 10x tokens, takes 20x longer, drains his weekly quota, and makes basic mistakes explicitly forbidden by the repo's guide — then claims the work is done, forcing manual diff review.
He suspects the regression is a business tactic to push users toward higher tiers or to filter out devs who don't review diffs, framing it as part of an aggressive token economy. This is one user's unverified experience, but it echoes widespread community complaints about model quality degradation.
More from coding & agent
- DeepSeek rumored to launch Super-Agent, an open-source coding agent to rival Claude Code — Polymarket · 2026-11-01
- Dev runs an agent that tracks CVEs and autonomously patches systems — aniketmaurya · 2026-11-01
- Agent autonomy needs a harness: layered state architecture survives a 10-day continuous run — burkov · 2026-11-01
- Sales agent builder: the plumbing breaks more often than the model — kushagra_zryth · 2026-11-01
- Developer reverse-engineers smart home protocols into a Raspberry Pi AR app — say "Lumos" to turn on lights — LinusEkenstam · 2026-11-01
- Open-source voice-agent leaderboard benchmarks turn detection and tool calls — ivan_digital · 2026-11-01