Been testing thinking levels on every model lately, so here’s DeepSeek V4-Flash 0731’s turn.
Same neon Tokyo prompt at off, low, high and max on 2 DGX Sparks. Off: 82s, 5.5K tokens Low: 7 min, 27.6K High: 13 min, 42.5K (after failing twice, 225K tokens of thinking with zero
You can use deepseek V4 pro for free and set it up in under 10 minutes
HERE’S HOW:
If you use Codex and you're stuck on GPT models only, this fixes that.
It's called codex-router, open source. drops other models straight into your normal Codex picker, right next to the GPT ones you already have. what it adds: - Grok, Kimi, Deepseek, Claude, all in the same
🇨🇳🇺🇸 The global AI race is becoming much closer with China closing the gap
US companies still produce many of the world’s most powerful models and possess vastly more computing capacity. But Chinese developers are rapidly closing the performance gap while competing
You have reached the end of the archive
All of deepseek