I just compared DeepSeek-v4.1-Flash against Opus 5
Gave same prompt to both models to build ocean sandbox using three.js DeepSeek-v4.1-Flash was almost 90% cheaper and 50% faster than Opus 5 💀 maybe the benchmarks are not broken.. do you think DeepSeek is better?
In v3.1.5 I gave the agent in Privacy AI one sentence on an iPhone 16 Pro Max, and it worked for the next 55 minutes
"Create a PPTX document to research and collect benchmarks among latest top LLM models including GPT-6, Fable-5.1, Grok-4.6, GLM-5.3 and DeepSeek-4.1-Flash" It
Not building today almost out of tokens, except DeepSeek so here is "Pretty fly for an AI guy"
Two cents against a dollar sixty.
Same design task, DeepSeek V4.1 Flash against Astra, scored by OpenDesign at 98 percent of Astra's result. The catch is real. On the wider benchmark it sits well behind, 39.1 against 60.3, so the hard jobs still go the expensive way. Most of my
You have reached the end of the archive
All of deepseek