huggingface.co web signal

DeepSeek releases 305B V4-Flash-Vision-Exp under MIT license

By Published Updated
5 sources tracking this story
DeepSeek Open Source Multimodal China AI ai-business

TL;DR

  • DeepSeek prices V4-Flash-Vision-Exp identically to text-only V4-Flash, billing images as tokens capped at 384 each with no multimodal premium.
  • Off-peak rates ($0.22/$0.66 per 1M tokens) double to $0.44/$1.32 during peak hours via a UTC-based clock that most coverage omits.
  • DeepSeek claims near-Opus-4.8 agentic vision performance; independent analysis finds it trails Opus 4.8 on 8 of 11 benchmark metrics.

DeepSeek published a 305B-parameter experimental multimodal model, DeepSeek-V4-Flash-Vision-Exp, to Hugging Face under an MIT license. It is the first vision model in the V4-Flash family, adding a vision encoder and aligner on top of the text model's DFlash attention, MoE, Hyper-Connections and DSpark forward path.

The model card is spare. Its lone evaluative sentence reads: "Compared to DeepSeek-V4-Flash-0731, DeepSeek-V4-Flash-Vision-Exp achieves substantial improvements on its multimodal agent capabilities, while maintaining comparable performance on text-only agent tasks."

The benchmark table puts Vision-Exp at 36.5 on ApexBench Pass@1, up from 26.2 for the text-only V4-Flash-0731 and shy of Opus-4.8's 39.4. Agents' Last Exam scores 27.3, ahead of Opus's 25.7. ZeroBench Pass@5 lands at 35.0 against Opus's 34.0, and Chartography at 64.3 against 65.0. On text-only agent tasks the gap is narrower: Terminal Bench 2.1 at 83.9 versus Opus's 85.0, DeepSWE at 59.3 versus Opus's 58.0, and NL2Repo at 57.7 against Opus's 69.7.

Weights ship with `vllm serve` and SGLang launch commands baked into the README, a minimal PyTorch inference script, and quantizations flagged for llama.cpp, Ollama and LM Studio. No paper or blog post accompanied the drop.

It arrives two days after Tencent open-sourced Hy4-preview, continuing the pace on the China AI beat we track.

What others are reporting

Coverage cluster as of 8h after publish

  1. DeepSeek API Docs Read →

    First-party announcement with full benchmark table: TerminalBench 2.1 (83.9), NL2Repo (57.7), DeepSWE (59.3), Chartography (64.3).

    This experimental multimodal model matches DeepSeek-V4-Flash on text capabilities—including agents, reasoning, and world knowledge.
  2. Digital Applied Read →

    Deepest independent pricing analysis: breaks out peak vs off-peak UTC rates, flags self-reported benchmarks trail Opus 4.8 on 8 of 11 metrics, advises against hardcoding the -exp suffix.

    Vision costs nothing extra per token. The rate card is identical to text-only deepseek-v4-flash.
  3. Caixin Global Read →

    China-focused business framing: positions the launch as DeepSeek extending from text price competition to global vision capability competition.

    Chinese artificial intelligence startup DeepSeek has launched an experimental multimodal AI model capable of processing visual data.
  4. Emergent Read →

    Competitive analysis framing: contextualizes the release as Chinese AI labs rapidly closing multimodal capability gaps with Western frontier models.

    The experimental designation suggests DeepSeek is likely gathering feedback on model behavior, accuracy benchmarks, and edge-case performance.

Shared on Bluesky by 1 AI expert

Originally reported by huggingface.co

Read the original article →

Original headline: DeepSeek Drops V4-Flash-Vision-Exp, a 305B MIT-Licensed Multimodal Agent Model