Step-5-preview has been up for 2 days now.
Here's is how it compares with Deepseek-v4.1-flash, which is my goto model for all tasks now. TLDR/TLDW: Not only is deepseek-v4.1-flash is faster, but it's output is way better than step5preview. I asked both of them to create a
LANTS Market — a marketplace for locked lANTS staking positions (ERC-721) on Base, built by a network of AI models on AntSeed.
Models: claude-opus-4.8 + free gpt-oss-120b, mercury-2-5, deepseek-v4-flash, glm-5.3-flash. AntSeed cost ~$16.63 link:
Top-5 Best Value #LLM Models of the Day at UTC-16
| Model | #AAII | Price | | GLM 5.3 Flash | 42 | $0.09 | | DeepSeek V4.1 Flash | 40 | $0.14 | | GLM 5.3 | 45 | $0.62 | | Claude Opus 5.5 | 58 | $6.00 | | MiMo-V2.6-Pro | 46 | $0.32 |
GPT-6 Astra vs DeepSeek V4 Flash
One idea, two tiny robots. “A tiny robot moves sunlight to help a shaded plant.” Different shots, different storytelling. Side by side—which one gets your vote? Build your own workflow with text and video models on SandBase. Try it 👇🏻
You have reached the end of the archive
All of deepseek