DeepSeek V4.1 Flash: 552B params at 420-507 tok/s, matches Claude Opus-5 on agent benchmarks
APPSO · wechat · 2026-09-12
DeepSeek released V4.1 Flash, unifying its chat, vision and expert models into one entry point. Despite doubling backbone params to 552B, it runs at 420-507 tok/s thanks to a redesigned architecture: a Causal Encoder-Decoder (CED) split that nearly halves long-input prefill compute, CSA2 cross-layer KV sharing with FP4 quantization, SWA Bounded Replay cutting persistent KV cache to 1/8 of the prior gen, and a 196B lookup-only Engram memory module. Per-token global KV cache is just 890 bytes — 1/437 of DeepSeek-V1. On agent benchmarks it hits 74.2% on DeepSWE v1.1 and beats Claude Opus-5 and GPT-5.6 Sol on Terminal-Bench 2.1 and AutomationBench, though gaps remain on the hardest tasks (Terminal-Bench 4.0: 31.2 vs Opus-5's 51.8). Post-training focused on task data engineering rather than new RL algorithms, and the report shows agent framework differences swing the same model's scores from 74.2 to 65.6.
Related event: DeepSeek launches V4.1-Flash: 552B backbone, 1M-token context(5 posts)→
More from coding & agent
- Does Claude finally make NixOS viable? AI-written configs vs. lingering pitfalls — jfischoff · 2026-10-31
- offense launches as an 'OpenRouter for agent security' — open source on GitHub — hamostaf04 · 2026-10-31
- 48-page deep dive: why modern LLM RL is secretly off-policy and 14 ways to fix it — cwolferesearch · 2026-10-31
- Serving agents over MCP is becoming consensus — and you can charge for access — pritisinghhhh · 2026-10-31
- Chasing the Karpathy Peak: mapping AI productivity's climb and the Slop Cliff past it — sirbayes · 2026-10-31
- Lauren Tan's agent playbook: set up rules and verification once, then scale to 200+ agents — anselm · 2026-10-31