Mistral AI Launches Mixtral 8x22B: A Powerhouse for Open Weights
Mistral AI publicly releases Mixtral 8x22B, a high-performance open-weight model challenging proprietary leaders with su…
🏷️ AI Benchmarks
39 articles about 'AI Benchmarks'
OpenAI Launches GPT-5 Turbo with Multimodal Reasoning
OpenAI unveils GPT-5 Turbo, featuring native multimodal reasoning across text, image, audio, and video inputs.
Claude 4 Opus Shatters Scientific Reasoning Benchmarks
Anthropic's Claude 4 Opus achieves record scores across multiple scientific reasoning benchmarks, outperforming GPT-4o a…
Gemini 2.5 Ultra Sets New Bar for Multimodal AI
Google's Gemini 2.5 Ultra achieves state-of-the-art results across major multimodal reasoning benchmarks, outperforming …
0% Pass Rate: New Benchmark Stumps All AI Models
The creators of SWE-Bench release a devastating new benchmark where Claude, GPT, and Gemini all score zero, exposing fun…
Google DeepMind Unveils Gemini 2.5 Ultra
Google DeepMind launches Gemini 2.5 Ultra, its most powerful AI model featuring native multimodal reasoning across text,…
Claude 4 Opus Breaks Science Reasoning Record
Anthropic's Claude 4 Opus achieves unprecedented scores on graduate-level science benchmarks, surpassing GPT-4o and Gemi…
Claude 4 Shatters Benchmark Records on Science Tasks
Anthropic's Claude 4 sets a new state-of-the-art score on graduate-level science benchmarks, outpacing GPT-4o and Gemini…
xAI Grok 3 Challenges GPT-4o in Math Benchmarks
Elon Musk's xAI releases Grok 3 with math reasoning scores rivaling OpenAI's GPT-4o, intensifying the LLM competition.
GPT-5 Reportedly Hits PhD-Level Reasoning
OpenAI's next-generation GPT-5 model reportedly achieves PhD-level reasoning on internal benchmarks, signaling a major l…
Gemini Ultra 2 Matches Humans on Grad Math Exams
Google DeepMind's Gemini Ultra 2 achieves human-level scores on graduate-level mathematics exams, marking a major milest…
Claude 4 Shatters Graduate-Level Science Benchmarks
Anthropic's Claude 4 sets new records on GPQA and other graduate-level science benchmarks, outpacing GPT-4o and Gemini U…