{"rewrite":{"id":"r_6e8d3d6efc982deafd15d76b","clusterId":"c_be1a599815709af1c917ec9e","slug":"deepreinforce-releases-open-coding-ai-ornith-1-0-top-model-beats-claude-opus-4-7","model":"deepseek-v4-flash","headline":"DeepReinforce Releases Open Coding AI Ornith-1.0, Top Model Beats Claude Opus 4.7","summary":"AI research organization DeepReinforce has released the Ornith-1.0 family of open models specialized for agent coding. The largest variant, Ornith-1.0-397B, outperforms Claude Opus 4.7 on multiple benchmarks. Four variants range from 9B to 397B parameters, with the smallest designed to run on edge devices. The models are distributed under the MIT License.","whyItMatters":"Ornith-1.0 treats agent scaffolds as learning targets rather than human-designed components, and its smallest model outperforms a much larger competitor, suggesting a shift in how efficient coding agents can be built.","webCardHtml":"\u003cp\u003eDeepReinforce introduced Ornith-1.0, a family of open AI models built for agent coding. The largest model, Ornith-1.0-397B, beat Claude Opus 4.7 across several benchmarks. Unlike typical agent systems where humans design the scaffold that controls behavior, Ornith-1.0 learns the scaffold as part of training while avoiding reward hacking. Four variants were announced: Ornith-1.0-9B, Ornith-1.0-31B, Ornith-1.0-35B, and Ornith-1.0-397B. The 9B and 31B are Dense models; the 35B and 397B use Mixture-of-Experts architecture. The 9B model outperformed the much larger Gemma 4 31B. Three variants are available now under the MIT License, with the 31B coming soon.\u003c/p\u003e\u003cul\u003e\u003cli\u003e\u003cstrong\u003eOrnith-1.0-9B\u003c/strong\u003e: Dense model, runs on edge devices, outperformed Gemma 4 31B\u003c/li\u003e\u003cli\u003e\u003cstrong\u003eOrnith-1.0-31B\u003c/strong\u003e: Dense model, scheduled for release soon\u003c/li\u003e\u003cli\u003e\u003cstrong\u003eOrnith-1.0-35B\u003c/strong\u003e: MoE model, outperformed Qwen3.6 35B and Gemma 4 31B\u003c/li\u003e\u003cli\u003e\u003cstrong\u003eOrnith-1.0-397B\u003c/strong\u003e: MoE model, comparable to Claude Opus 4.7, outperformed it on multiple benchmarks\u003c/li\u003e\u003c/ul\u003e","blueskyPost":"DeepReinforce's Ornith-1.0-397B beats Claude Opus 4.7 on benchmarks while being open-source under MIT License. The real competition isn't top scores but whether open models can match closed-source reliability in production.","twitterPost":"Ornith-1.0-397B beats Claude Opus 4.7 on benchmarks. The MIT License makes this a direct challenge to closed-source dominance.","threadsPost":"DeepReinforce's Ornith-1.0 family includes a 397B model that outperforms Claude Opus 4.7 on benchmarks. All variants are MIT-licensed. The small 9B variant runs on edge devices. The real test is whether open weights can match closed-source reliability in actual coding workflows.","newsletterBlurb":"DeepReinforce released the Ornith-1.0 family of open AI models specialized for agent coding. The largest variant, Ornith-1.0-397B, outperformed Claude Opus 4.7 on multiple benchmarks. Four variants range from 9B to 397B parameters, with the smallest designed for edge devices. The models are distributed under the MIT License.","attributionJson":"[{\"source\":\"GIGAZINE\",\"url\":\"https://gigazine.net/news/20260629-ornith-agentic-coding-ai/\",\"title\":\"Coding AI Model \\\"Ornith-1.0\\\" with Performance Equivalent to Claude Opus 4.7 Appears, Lineup Includes Small Models That Run Locally\"}]","lintFlagsJson":null,"lintHits":0,"costUsd":0,"inputTokens":3968,"outputTokens":822,"status":"published","repairAttempts":0,"nextRepairAt":null,"factsAttemptedAt":1782709010,"createdAt":"2026-06-29T04:50:15.000Z","publishedAt":"2026-06-29T04:54:50.000Z","updatedAt":"2026-06-29T04:54:50.000Z"},"cluster":{"id":"c_be1a599815709af1c917ec9e","canonicalTitle":"Claude Opus 4.7と同等性能のコーディングAIモデル「Ornith-1.0」が登場、ローカルで動作する小型モデルもラインナップ","representativeArticleId":"a_ffed5276a4acb82ea92bb6fc","sourceCount":1,"writtenSourceCount":1,"writeAttempts":0,"isSolo":true,"entitiesJson":"{\"anime_titles\":[],\"manga_titles\":[],\"work_titles\":[],\"studios\":[],\"people\":[],\"type\":\"news\",\"domain\":\"other\",\"is_roundup\":false}","contentType":"news","status":"published","firstSeenAt":"2026-06-29T04:16:00.000Z","lastSeenAt":"2026-06-29T04:16:00.000Z","updatedAt":"2026-06-29T04:54:50.000Z"},"attribution":[{"source":"GIGAZINE","url":"https://gigazine.net/news/20260629-ornith-agentic-coding-ai/","title":"Claude Opus 4.7と同等性能のコーディングAIモデル「Ornith-1.0」が登場、ローカルで動作する小型モデルもラインナップ"}],"entities":{"anime_titles":[],"manga_titles":[],"work_titles":[],"studios":[],"people":[],"type":"news","domain":"other","is_roundup":false},"keyFacts":["DeepReinforce released the Ornith-1.0 family of open AI models for agent coding under the MIT License.","The largest variant, Ornith-1.0-397B, outperformed Claude Opus 4.7 on multiple benchmarks.","Four variants range from 9B to 397B parameters: Ornith-1.0-9B, Ornith-1.0-31B, Ornith-1.0-35B, and Ornith-1.0-397B.","The 9B model runs on edge devices and outperformed the larger Gemma 4 31B.","Ornith-1.0 learns the agent scaffold as part of training, unlike typical systems where humans design the scaffold."]}