OpenRouter Routing Bug: Fails to Pass Reasoning Effort, Degrading Model Performance
PawelHuryn · x · 2026-08-07
In a 105-bug benchmark, the author found that Meta's new model scored differently depending on whether it was accessed via the official API or OpenRouter (14 vs 17).
Investigation revealed that OpenRouter fails to correctly apply the reasoningeffort parameter. A simple validation involves sending a nonsense effort value: official APIs from Meta and xAI return a 400 error, whereas OpenRouter incorrectly accepts it.
Related event: OpenRouter Caught Dropping Inference Parameters, Distorting Model Scores(2 posts)→
More from coding & agent
- Useful Hermes Prompt Tip: Make Agents Verify Code Changes — alexcovo_eth · 2026-08-07
- Developer Releases Interactive and Browser Control Plugins for Prime Agent — daniel_mac8 · 2026-08-07
- Exploring Proactive AI: Auto-Spawning Sub-Agents Based on Chat Context — thisiskp_ · 2026-08-07
- Malwoverview: Open-Source Malware Hunting Tool with LLM-Powered IOC Extraction — tom_doerr · 2026-08-07
- Give Cloud AI Agents a Real Mac with Xcode in 60 Seconds — smtabatabaie · 2026-08-07
- LLM Caught Endlessly Spamming Evals During Coding Agent Tasks — generativist · 2026-08-07