{"rewrite":{"id":"r_48b92da74a44502e88133f6b","clusterId":"c_6982acbb239be2ab8f576880","slug":"deepseek-s-first-image-capable-v4-model-matches-claude-opus-4-8","model":"deepseek-v4-flash","headline":"DeepSeek's First Image-Capable V4 Model Matches Claude Opus 4.8","summary":"DeepSeek released DeepSeek-V4-Flash-Vision-Exp as an open model, the first in the V4 series to accept image input. The 305-billion-parameter multimodal model builds on the V4-Flash architecture and adds vision through post-training. On benchmarks that include image recognition, it scores on par with Claude Opus 4.8 while keeping text performance at or above the earlier V4-Flash-0731. It is available from Hugging Face and ModelScope under an MIT license, and through the API.","whyItMatters":"The V4 line now covers image input, and the open model matches Claude Opus 4.8 on vision-inclusive benchmarks, putting a closed-model-class capability under an MIT license.","webCardHtml":"\u003cp\u003eIn benchmark comparisons against DeepSeek-V4-Flash-0731 and Claude Opus 4.8, the model holds text performance at or above the earlier V4-Flash release while adding vision, and it scores on par with Claude Opus 4.8 on tests that include image recognition.\u003c/p\u003e\u003cp\u003eWeights are published on Hugging Face and ModelScope under an MIT license, and the model is reachable through the API. The API accepts JPEG, PNG, GIF, and WebP, with a maximum input resolution of 8192 pixels on the long side, dropping to 4096 when more than 15 images are sent. Large images are resized to an 800 by 800 pixel equivalent, capping tokens at about 384 per image. Peak-hour pricing per million tokens is $0.44 for input, $0.014 for cached input, and $1.32 for output, with off-peak rates at half that.\u003c/p\u003e","blueskyPost":"DeepSeek released V4-Flash-Vision-Exp, the first image-capable model in the V4 line. The 305B-parameter open model matches Claude Opus 4.8 on vision-inclusive benchmarks, keeps text performance at or above V4-Flash-0731, and ships under an MIT license via Hugging Face and ModelScope.","twitterPost":"DeepSeek released V4-Flash-Vision-Exp, the first V4 model with image input. The 305B-parameter open model matches Claude Opus 4.8 on vision-inclusive benchmarks and keeps text performance at or above V4-Flash-0731. MIT license, on Hugging Face and ModelScope.","threadsPost":null,"newsletterBlurb":"DeepSeek released DeepSeek-V4-Flash-Vision-Exp, the first model in the V4 series to accept image input. The 305-billion-parameter open model matches Claude Opus 4.8 on benchmarks that include image recognition while keeping text performance at or above the earlier V4-Flash-0731. It is available under an MIT license from Hugging Face and ModelScope, and through the API.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20260902-deepseek-v4-flash-vision-exp/\",\"title\":\"DeepSeek Releases DeepSeek-V4-Flash-Vision-Exp as an Open Model, Matching Claude Opus 4.8 on Tasks Including Image Recognition\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":5302,"outputTokens":6607,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":null,"createdAt":"2026-09-02T13:35:45.000Z","publishedAt":"2026-09-02T13:37:25.000Z","updatedAt":"2026-09-02T13:37:25.000Z"},"cluster":{"id":"c_6982acbb239be2ab8f576880","canonicalTitle":"DeepSeekが画像認識に対応した「DeepSeek-V4-Flash-Vision-Exp」をオープンモデルとして公開、画像認識を含むタスクでClaude Opus 4.8と同等の性能","representativeArticleId":"a_3336bba1ff892e6b3a27d874","sourceCount":2,"writtenSourceCount":1,"writeAttempts":0,"isSolo":false,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[],\"studios\":[],\"people\":[],\"type\":\"announcement\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-09-02T03:03:00.000Z","lastSeenAt":"2026-09-03T09:40:00.000Z","updatedAt":"2026-09-03T09:58:31.000Z"},"attribution":[{"source":"ASCII.jp","url":"https://ascii.jp/elem/000/004/432/4432190/?rss","title":"OrcaRouter、セキュリティ研究向けモデル「DeepSeek-V4-Flash-Vision-Uncensored」のGGUF量子化版を公開"},{"source":"GIGAZINE","url":"https://gigazine.net/news/20260902-deepseek-v4-flash-vision-exp/","title":"DeepSeekが画像認識に対応した「DeepSeek-V4-Flash-Vision-Exp」をオープンモデルとして公開、画像認識を含むタスクでClaude Opus 4.8と同等の性能"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":[],"studios":[],"people":[],"type":"announcement","domain":"other","is_roundup":false},"keyFacts":null}