Breakingpricing2m ago
Anthropic drops Claude Sonnet input price by 30%
Effective immediately. Significant for high-volume inference workloads.
model1h ago
Gemini 2.5 Flash posts new MMLU high — beats GPT-4o on reasoning
89.4 on MMLU Pro. Benchmark details now live in the tracker.
tool3h ago
Cursor ships background agents — async code tasks while you sleep
Agents push PRs autonomously. Solo dev game-changer, currently in beta.
model5h ago
Mistral releases Codestral 2 with 256k context window
Open weights, commercial licence. Strong on HumanEval+ and SWE-Bench.