{"slug": "show-hn-i-found-a-more-accurate-upcoming-llm-benchmarks-leaderboard", "title": "Show HN: I found a more accurate upcoming LLM Benchmarks leaderboard", "summary": "An independent LLM testing leaderboard published on Hacker News ranks Anthropic's Claude Opus 5.5 first on a verification score of 85.0, the share of 80 questions answered correctly, ahead of Google's Gemini 4 Argon at 84.8 and OpenAI's GPT-5.3-Codex and Google's Gemini 3.8 Flash, both at 82.5. The leaderboard also compares models on median output tokens per second, average billed USD with reasoning included, and median seconds to first token, and breaks scores down across Science (30 questions), Engineering (15), Medicine (11), Economics (9), Law (6) and Humanities & social science (9). The table covers 20 models from Anthropic, Google, OpenAI, SpaceXAI, Meta, Qwen, MoonshotAI and DeepSeek, with Claude Opus 5.5 marked as having no cheaper model scoring as high.", "body_md": "## Verification score\n\nShare of 80 questions answered correctly · higher is better\n\n Google Anthropic OpenAI SpaceXAI DeepSeek Meta Qwen MoonshotAI Other labs\n\nIndependent AI model testing\n\nUnbiased comparisons of the leading LLMs on accuracy, speed, latency and real cost, so you can choose the right model for your use case.\n\nFilled dots: no cheaper model scores as high · Claude Opus 5.5 has the top score\n\nMedian output tokens per second · higher is better\n\nAverage billed USD, reasoning included · lower is better\n\nMedian seconds to the first token · lower is better\n\nQuestions answered correctly in each category · categories hold 6 to 30 questions, so read single cells with care\n\n| Model | Score | Science30 questions | Engineering15 questions | Medicine11 questions | Economics9 questions | Law6 questions | Humanities & social science9 questions | \n|---|---|---|---|---|---|---|---|\n| [Claude Opus 5.5](https://www.llmverifications.com/models/claude-opus-5-5) Anthropic | 85.0 | 28/30 | 12/15 | 7/11 | 8/9 | 5/6 | 8/9 | \n| [Gemini 4 Argon](https://www.llmverifications.com/models/gemini-4-argon) Google | 84.8 | 25/30 | 13/14 | 7/11 | 8/9 | 6/6 | 8/9 | \n| [GPT-5.3-Codex](https://www.llmverifications.com/models/gpt-5-3-codex) OpenAI | 82.5 | 28/30 | 11/15 | 7/11 | 8/9 | 4/6 | 8/9 | \n| [Gemini 3.8 Flash](https://www.llmverifications.com/models/gemini-3-8-flash) Google | 82.5 | 26/30 | 12/15 | 6/11 | 8/9 | 6/6 | 8/9 | \n| [Claude Fable 5.1](https://www.llmverifications.com/models/claude-fable-5-1) Anthropic | 81.2 | 25/30 | 13/15 | 7/11 | 7/9 | 5/6 | 8/9 | \n| [Grok 4.7](https://www.llmverifications.com/models/grok-4-7) SpaceXAI | 80.0 | 27/30 | 11/15 | 6/11 | 8/9 | 4/6 | 8/9 | \n| [Muse Wonder](https://www.llmverifications.com/models/muse-wonder) Meta | 78.8 | 26/30 | 11/15 | 5/11 | 8/9 | 4/6 | 9/9 | \n| [Claude Sonnet 5.5](https://www.llmverifications.com/models/claude-sonnet-5-5) Anthropic | 78.8 | 24/30 | 13/15 | 7/11 | 8/9 | 3/6 | 8/9 | \n| [Gemini 3.1 Pro Preview](https://www.llmverifications.com/models/gemini-3-1-pro-preview) Google | 77.5 | 26/30 | 10/15 | 7/11 | 8/9 | 3/6 | 8/9 | \n| [Muse Spark Open](https://www.llmverifications.com/models/muse-spark-open) Meta | 77.5 | 25/30 | 11/15 | 6/11 | 8/9 | 4/6 | 8/9 | \n| [Kimi K3.1 Pro](https://www.llmverifications.com/models/kimi-k3-1-pro) MoonshotAI | 77.5 | 26/30 | 11/15 | 6/11 | 8/9 | 3/6 | 8/9 | \n| [GPT-6 Astra](https://www.llmverifications.com/models/gpt-6-astra) OpenAI | 76.2 | 26/30 | 10/15 | 6/11 | 8/9 | 3/6 | 8/9 | \n| [GPT-6.1 Terra Pro](https://www.llmverifications.com/models/gpt-6-1-terra-pro) OpenAI | 76.2 | 26/30 | 10/15 | 6/11 | 8/9 | 3/6 | 8/9 | \n| [GPT-6.1 Sol](https://www.llmverifications.com/models/gpt-6-1-sol) OpenAI | 76.2 | 27/30 | 10/15 | 5/11 | 8/9 | 3/6 | 8/9 | \n| [GPT-5.6 Terra](https://www.llmverifications.com/models/gpt-5-6-terra) OpenAI | 75.0 | 26/30 | 11/15 | 4/11 | 7/9 | 4/6 | 8/9 | \n| [Gemini 3.5 Flash Lite](https://www.llmverifications.com/models/gemini-3-5-flash-lite) Google | 75.0 | 23/30 | 10/15 | 7/11 | 7/9 | 4/6 | 9/9 | \n| [Qwen3.8 Max (0902)](https://www.llmverifications.com/models/qwen3-8-max-0902) Qwen | 75.0 | 24/30 | 9/15 | 6/11 | 8/9 | 4/6 | 9/9 | \n| [GPT-6.1 Terra](https://www.llmverifications.com/models/gpt-6-1-terra) OpenAI | 73.8 | 26/30 | 9/15 | 6/11 | 7/9 | 3/6 | 8/9 | \n| [DeepSeek V4.1 Flash 1017](https://www.llmverifications.com/models/deepseek-v4-1-flash-1017) DeepSeek | 73.8 | 26/30 | 9/15 | 4/11 | 7/9 | 4/6 | 9/9 | \n| [GPT-6 Sol](https://www.llmverifications.com/models/gpt-6-sol) OpenAI | 73.8 | 27/30 | 8/15 | 5/11 | 7/9 | 4/6 | 8/9 | \n\nComputed from this run by the rule in each label", "url": "https://wpnews.pro/news/show-hn-i-found-a-more-accurate-upcoming-llm-benchmarks-leaderboard", "canonical_source": "https://www.llmverifications.com/", "published_at": "2026-10-27 18:25:14+00:00", "updated_at": "2026-10-27 18:41:49.099988+00:00", "lang": "en", "topics": ["large-language-models", "artificial-intelligence", "ai-research"], "entities": ["Anthropic", "Claude Opus 5.5", "Google", "Gemini 4 Argon", "OpenAI", "GPT-5.3-Codex", "Gemini 3.8 Flash", "Meta"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/show-hn-i-found-a-more-accurate-upcoming-llm-benchmarks-leaderboard", "markdown": "https://wpnews.pro/news/show-hn-i-found-a-more-accurate-upcoming-llm-benchmarks-leaderboard.md", "text": "https://wpnews.pro/news/show-hn-i-found-a-more-accurate-upcoming-llm-benchmarks-leaderboard.txt", "jsonld": "https://wpnews.pro/news/show-hn-i-found-a-more-accurate-upcoming-llm-benchmarks-leaderboard.jsonld"}}