DeepSeek-V4.1-Flash tops Agent Arena open models, ranks #11 overall at $0.20/task
vista8 · x · 2026-10-20
DeepSeek-V4.1-Flash-20261017 (Max) is now the #1 open model on the Agent Arena leaderboard, landing at #11 overall with a +5.1% net improvement across 7,500+ real-world sessions.
- #8 in Confirmed Success (+11.5%, #1 open)
- #11 in Praise vs. Complaint (+5.9%, #1 open)
- #12 in Bash Reliability (+5.4%, #1 open)
- At a $0.20 median cost per task, it delivers 37% of GPT-6.1 Sol's improvement for 27% of the cost, reshaping the Pareto frontier — every model above it is proprietary and pricier
DeepSeek also open-sourced its internal inference framework DeepInfer the same day, underscoring its infra strength.
More from Infra
- Chrome 155 ships JPEG XL: 30-50% smaller than JPEG with lossless JPEG transcode — banteg · 2026-10-20
- Polymarket: DeepSeek releases open-source tool to sharply cut AI inference compute costs — Polymarket · 2026-10-20
- Together, Fireworks, Prime Intellect and Modal are all converging on inference + RL training — NielsRogge · 2026-10-20
- NVIDIA GTC kicks off in Berlin with Jensen Huang keynote set for tomorrow — NVIDIAAI · 2026-10-20
- Hyperdrives CEO interview: cheaper power-dense electric motors via cooling breakthrough — MarvinTBaumann · 2026-10-20
- MiniMax H3 video generation runs on a single RTX 3060 12GB via GGUF quantization — Ok-Inevitable8391 · 2026-10-20