Phind

The 91 comments (all HackerNews, from Phind model-launch threads) center on the claim that the Phind model 'beats GPT-4 at coding,' and the crowd verdict is mixed. Search is the clearest win: one user who spent hours hunting an obscure wave-solderable 8-pin GPU-power connector across major electronics distributors says Phind 'found it in seconds' and calls it 'a much better search engine than google/kagi/bing.' A self-taught IT admin automating Python/PowerShell found Phind 'truly helpful' vs ChatGPT. Head-to-head model tests split: on a vague distributed-job-queue design, Phind 'was able, even eager, to recommend specific libraries' matching the user's own research (GPT-4 'takes some coaxing') and ran faster; one tester called open-weight Phind-70B 'really good... maybe even GPT-4 level' on a TypeScript/web3 smart-contract hook. But on a concrete PostgreSQL teaser-update task, another user reports 'GPT4 still outperforms.' Hallucination recurs: Phind confidently claimed Cisco LISP supports map-server database replication ('it doesn't... it basically said "yeah it does, here's how"'), and a Haskell user said it 'tried to bullshit me about validity of its answer' citing an arXiv paper that didn't support it. One user found Search mode 'heavily diluted' an expert IoT device-authentication query with irrelevant methods (multi-factor auth for machine-to-machine), preferring Chat mode; another hit total code-generation failure on a LeetCode-style max-profit prompt after enabling an option. Credibility friction: a top comment calls the 'beats GPT-4' claim unsupported and 'quite deceptive' since the only linked benchmark excludes GPT-4; several distrust coding benchmarks generally (HumanEval contamination) and demand tasks like real GitLab CI bootstrapping. Pricing draws specific fire: the '8x difference in price-per-search between Plus and Pro' reads as either 'huge margins' or an unsustainable subsidy, and daily search limits are disliked because 'usage isn't uniform day-to-day.' Minor gripe: dark-mode text too heavy/bright for long reading. Some comments are general LLM-industry meta (investor money, sampling research), not Phind-specific.

No aggregate ratings were found for this product during the last harvest.