{"slug": "claude-opus-is-10x-faster-than-openai-gpt-5-at-non-streaming-completions", "title": "Claude Opus is 10x faster than OpenAI GPT 5 at non-streaming completions", "summary": "An independent developer reports that Anthropic's Claude Opus 4.8 is nearly 10 times faster than OpenAI's GPT-5 at non-streaming completions, based on side-by-side testing of blog comment rewrites. The developer also found that OpenAI's GPT-5-mini is both cheaper and faster than GPT-5, with input token pricing at $0.75 versus $2.50 for GPT-5.4, and is moving away from the litellm wrapper to use native SDKs from Anthropic and OpenAI due to security concerns.", "body_md": "This picture summarizes it well:\n\nHere on my blog, for [this popular blog post](/plog/blogitem-040601-1) I get a lot of comments. 28k blog comments over the years. Some of them are terribly written and hard to understand, so I let AI suggest a rewrite. That code that sends the blog post comment to AI, I actually fire off three times: once with OpenAI `gpt-5`\n\n, once with OpenAI `gpt-5-mini`\n\n, and once with Claude `claude-opus-4.8`\n\n. I use my human eyes and judgement to evaluate the results, and I can tell you they do equally well. Only the slightest differences.\n\nThe surprising thing is how amazingly slow OpenAI's `gpt-5`\n\nis! It's nearly 10x slower than `claude-opus-4.8`\n\n. What's up with that!?\n\nIt's also clear that the latency difference between `gpt-5-mini`\n\nand `gpt-5`\n\nis significant. At the time of writing, the input token price difference between `gpt-5.4`\n\nand `gpt-5.4-mini`\n\nis $2.50 compared to $0.75! That's a 3x difference.\n\n[Conclusion](#conclusion)\n\n-\nIf you're constructing a prompt the API, use Claude.\n\n-\nIf you have to use OpenAI, consider the\n\n`mini`\n\nmodel because it's*both*cheaper*and*faster.\n\n[Bonus](#bonus)\n\nBefore I added Claude, I used to use `litellm`\n\nto wrap OpenAI's models. The code looks like this:\n\n```\nresponse = litellm.completion(\n    model=\"openai-gpt-5\",\n    api_key=settings.OPENAI_API_KEY,\n    messages=my_prompt_messages,\n)\n```\n\nUnlike, if you use the native OpenAI Python SDK the invocation looks like this:\n\n```\nclient = openai.OpenAI(api_key=settings.OPENAI_API_KEY)\nresponse = client.responses.create(\n    model=\"gpt-5,\n    input=my_prompt_messages,\n)\n```\n\nI measured the difference, in speed, where I compare using the OpenAI SDK versus the `litellm`\n\nwrapper and the difference looks like this:\n\nGranted, in June I \"only\" did a bit over 30 of these calls, but strangely there's a difference!\n\nI don't have the intricate knowledge to understand why the `litellm`\n\nmakes the total time different from using the SDK. (not sure I care either!)\n\nEither way, I'm moving away from `litellm`\n\nand only use the SDKs provided by Claude and OpenAI. Feels safer given the CVEs we've seen this year on `litellm`\n\n.", "url": "https://wpnews.pro/news/claude-opus-is-10x-faster-than-openai-gpt-5-at-non-streaming-completions", "canonical_source": "http://www.peterbe.com/plog/claude-opus-is-10x-faster-than-openai-gpt-5-at-non-streaming-completions", "published_at": "2026-07-24 15:48:18+00:00", "updated_at": "2026-07-28 21:28:21.576354+00:00", "lang": "en", "topics": ["large-language-models", "ai-products", "developer-tools"], "entities": ["Anthropic", "OpenAI", "Claude Opus 4.8", "GPT-5", "GPT-5-mini", "litellm"], "alternates": {"html": "https://wpnews.pro/news/claude-opus-is-10x-faster-than-openai-gpt-5-at-non-streaming-completions", "markdown": "https://wpnews.pro/news/claude-opus-is-10x-faster-than-openai-gpt-5-at-non-streaming-completions.md", "text": "https://wpnews.pro/news/claude-opus-is-10x-faster-than-openai-gpt-5-at-non-streaming-completions.txt", "jsonld": "https://wpnews.pro/news/claude-opus-is-10x-faster-than-openai-gpt-5-at-non-streaming-completions.jsonld"}}