After 7B+ tokens, developer says DeepSeek V4.1 Flash is his default workhorse
gaganghotra_ · x · 2026-09-19
- Developer @btsouth reports pushing 7 billion+ tokens through DeepSeek V4.1 Flash and concludes it's not a budget fallback but his default workhorse.
- He routes coding, debugging, refactors, repo exploration, test/fix loops, and multiple parallel agents to DeepSeek; his usual premium models (Astra and Fable) have become nearly irrelevant for daily work.
- Despite its very low price, he says the model doesn't feel benchmaxxed — if anything, the benchmark sheet undersells it, and he still hasn't hit a task that forces him back to pricier models.
Related event: DeepSeek Flash Becomes Default Workhorse, Cutting AI Costs 90%(2 posts)→
More from Models
- Dev suggests AI companies should refund API credits on agent refusals — GabGarrett · 2026-09-19
- Dev claims Codex users are 'getting scammed' over usage terms — AIFlow_ML · 2026-09-19
- Step 5 Preview Quietly Debuts on AA: Score 44, 1M Context, $1/1M Input — teortaxesTex · 2026-09-19
- Developers say Codex is 'not sustainable' under current usage limits — AIFlow_ML · 2026-09-19
- Cognition ships SWE-2 coding model: 1 point behind Fable 5.1 at 64% less cost — AxSaucedo · 2026-09-19
- Jev Hits 36M Views in 2 Days, Community Ships 6 Open Clones — Latent Space · 2026-09-19