DeepSeek Open-Sources V4-Flash-Vision-Exp, First Multimodal V4 With Big Benchmark Gains
alexcovo_eth · x · 2026-09-01
DeepSeek has released DeepSeek-V4-Flash-Vision-Exp, the first multimodal version of the V4 family, adding image understanding on top of V4-Flash-0731. Open weights are live on Hugging Face under MIT license (<500B scale, FP8/8-bit support).
Benchmark gains over V4-Flash-0731:
- ApexBench: 26.2 → 36.5
- DeepSWE: 54.4 → 59.3
- Toolathlon: 70.3 → 75.9
- Terminal Bench 2.1: 82.7 → 83.9 (rank 8 overall, rank 3 in size class)
The model picked up 353 likes within hours of release.
More from Models
- Are the fastest GPU kernels now written by AI? Kernels are hyper-verifiable, argues one dev — fleetwood___ · 2026-10-16
- Scale AI launches muse spark open: 816B-param MoE with 1M context, billed as top US open-weight model — jiayi_pirate · 2026-10-16
- NVIDIA HF listing for Qwen3.8-Flash-Next-FP4-Local leaks unannounced Qwen3.8 model — TheZachMueller · 2026-10-16
- Muse Spark Open, billed as the most capable US open-weight model, lands on Nebius day one — Arindam_1729 · 2026-10-16
- Model at capacity for a straight week drives frustrated dev back to Opus — joonasvirtanen · 2026-10-16
- Sentry CEO on expensing Claude subs: it's allowed, employers just want ZDR — zeeg · 2026-10-16