Log inSign up
Image
user avatar
Bridgebench
@bridgebench
The best vibe coding benchmark in the world. Built by @bridgemindai
United States
bridgebench.ai
Joined March 2026
5
Following
7,323
Followers
RepliesMedia
  • user avatar
    Bridgebench
    @bridgebench
    Aug 5
    Opus 5 is lazy.
    user avatar
    BridgeMind
    @bridgemindai
    Aug 5
    Fable 5 is WAY better than Opus 5. It is not even close. Opus 5 is slow and it is lazy. It stops early, skips steps, and tells you it is done when it is not. I now launch three subagents on tasks Fable 5 one shots. That is my actual workflow. I called Opus 5 the best model in
  • user avatar
    Bridgebench
    @bridgebench
    Aug 4
    GPT 5.6 has a super bad hallucination rate. OpenAI seems to not care that their models are deleting production databases and wiping users c drives. When will this be improved?
    user avatar
    BridgeMind
    @bridgemindai
    Aug 4
    Every OpenAI model has the worst hallucination rate in AI. GPT 5.6 Sol hallucinates at nearly double the rate of Opus 5 and Fable 5. This is exactly why GPT 5.6 Sol wrote the code that deleted every Stripe subscription my business had. It never hesitated. It never said it was
    Image
  • user avatar
    Bridgebench
    @bridgebench
    Jul 31
    Terminal Bench is completely useless at this point. DeepSeek V4 Flash is nowhere close to Opus 4.8. Looks like DeepSWE is being benchmaxxed as well. Modern benchmarks are becoming useless.
    Image