DeepSeek V4-Flash Silently Upgraded: Terminal-Bench Score Jumps 25.8 Points
alejandroll10 · x · 2026-08-01
DeepSeek has silently rolled out a V4-Flash upgrade accessible via their API.
On the Terminal-Bench benchmark, the model achieved a score of 82.7, marking a massive 25.8-point leap from its April preview score of 56.9. Open weights for this version are expected to be released shortly.
Related event: DeepSeek V4-Flash API Public Beta Launches with Major Agent Upgrades(5 posts)→
More from Models
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- Google won't publish Argon 4 limits in AI Pro—users compare experiences — Consistent-Text-2662 · 2026-10-16
- Tencent releases Hy-Memory-30B-A3B, a 3B-active memory-focused model claiming to beat Opus — victormustar · 2026-10-16
- Reddit user puzzled why so many prefer Claude over 'much better' ChatGPT — -ButterflyEffect- · 2026-10-16
- User rants GPT-6.1 Sol is painfully slow and buggy, says Claude is 7x faster — karmay007 · 2026-10-16
- $20-plan users have no idea how far AI agents have come in 10 months — realmeetjames · 2026-10-16