{"id":93299,"topic":"ai","source":"opinion.caixin.com","title":"大模型产业的开源与闭源之争 - opinion.caixin.com","url":"https://opinion.caixin.com/2026-09-30/102489915.html","url_hash":"c61a6b598fa077ce812172274fccddfb271d9c66","author":"","summary":"<a href=\"https://news.google.com/rss/articles/CBMiYkFVX3lxTE96bDlfU1ZvVWlSVXpYV21FY3hOanBBaFdIZExQXzZ4cmdpZWZxUk1PNmVNQVJ2MTVnV3lEVlZ4NUhTWmpuZUQzT2JQd19tamUxT0xFZWtPbXpMeVRKNmJKVzdR?oc=5\" target=\"_blank\">大模型产业的开源与闭源之争</a>&nbsp;&nbsp;<font color=\"#6f6f6f\">opinion.caixin.com</font>","content":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","image_url":null,"lang":"zh","published_at":"2026-09-30T03:55:00+00:00","fetched_at":"2026-09-30T05:15:08+00:00","status":"read","starred":0,"extract_state":"ok","summary_auto":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","cluster_id":null,"extract_retries":0,"extract_error":null,"contract_version":"news_item.v1","format_contract_version":"news_item_formats.v1","dedup_url":"https://opinion.caixin.com/2026-09-30/102489915.html","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 365 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":365,"summary_length":365,"usable_text_length":365,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":365,"summary_length":365}},"news_item":{"id":93299,"canonical_url":"https://opinion.caixin.com/2026-09-30/102489915.html","source_url":"https://opinion.caixin.com/2026-09-30/102489915.html","title":"大模型产业的开源与闭源之争 - opinion.caixin.com","source_name":"opinion.caixin.com","author":null,"published_at":"2026-09-30T03:55:00+00:00","locale":"zh","topic":"ai","tags":[],"rss_summary":"<a href=\"https://news.google.com/rss/articles/CBMiYkFVX3lxTE96bDlfU1ZvVWlSVXpYV21FY3hOanBBaFdIZExQXzZ4cmdpZWZxUk1PNmVNQVJ2MTVnV3lEVlZ4NUhTWmpuZUQzT2JQd19tamUxT0xFZWtPbXpMeVRKNmJKVzdR?oc=5\" target=\"_blank\">大模型产业的开源与闭源之争</a>&nbsp;&nbsp;<font color=\"#6f6f6f\">opinion.caixin.com</font>","full_text":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","excerpt":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","extraction":{"state":"ok","confidence":0.9,"error":null,"explanation":"High confidence: full text extraction produced 365 characters.","diagnostics_url":"/api/diagnose?url=https%3A//opinion.caixin.com/2026-09-30/102489915.html","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 365 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":365,"summary_length":365,"usable_text_length":365,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":365,"summary_length":365}}},"display_formats":["compact","card","full","digest_section","json"]},"daily_stack_record":{"title":"大模型产业的开源与闭源之争 - opinion.caixin.com","url":"https://opinion.caixin.com/2026-09-30/102489915.html","summary":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","source":"opinion.caixin.com","date":"2026-09-30T03:55:00+00:00","content":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","confidence":0.9,"diagnostics_url":"/api/diagnose?url=https%3A//opinion.caixin.com/2026-09-30/102489915.html","quality_bucket":"high","failure_kind":"none","retryable":false,"quality_reason":"High confidence: full text extraction produced 365 characters.","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 365 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":365,"summary_length":365,"usable_text_length":365,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":365,"summary_length":365}},"tags":[]},"fallback_formats":["markdown","json","html"],"actions":{"read":"/item/93299","export_markdown":"/api/items/93299/export?format=markdown","export_json":"/api/items/93299/export?format=json","diagnose":"/api/diagnose?url=https%3A//opinion.caixin.com/2026-09-30/102489915.html"},"formats":{"full":{"id":93299,"title":"大模型产业的开源与闭源之争 - opinion.caixin.com","url":"https://opinion.caixin.com/2026-09-30/102489915.html","source":"opinion.caixin.com","author":null,"published_at":"2026-09-30T03:55:00+00:00","locale":"zh","topic":"ai","tags":[],"excerpt":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","full_text":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","reading_time_min":1,"extraction":{"state":"ok","confidence":0.9,"error":null,"explanation":"High confidence: full text extraction produced 365 characters.","diagnostics_url":"/api/diagnose?url=https%3A//opinion.caixin.com/2026-09-30/102489915.html","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 365 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":365,"summary_length":365,"usable_text_length":365,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":365,"summary_length":365}}},"quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 365 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":365,"summary_length":365,"usable_text_length":365,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":365,"summary_length":365}},"actions":{"read":"/item/93299","export_markdown":"/api/items/93299/export?format=markdown","export_json":"/api/items/93299/export?format=json","diagnose":"/api/diagnose?url=https%3A//opinion.caixin.com/2026-09-30/102489915.html"}},"digest":{"id":93299,"title":"大模型产业的开源与闭源之争 - opinion.caixin.com","url":"https://opinion.caixin.com/2026-09-30/102489915.html","source":"opinion.caixin.com","topic":"ai","published_at":"2026-09-30T03:55:00+00:00","excerpt":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。 2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。 但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse…","quality_bucket":"high","quality_reason":"High confidence: full text extraction produced 365 characters.","reading_time_min":1,"cluster_id":null},"card":{"display_title":"大模型产业的开源与闭源之争 - opinion.caixin.com","subtitle":"opinion.caixin.com · 2026-09-30","summary":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。 2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。…","badges":["quality:high"],"links":{"read":"/item/93299","original":"https://opinion.caixin.com/2026-09-30/102489915.html","diagnose":"/api/diagnose?url=https%3A//opinion.caixin.com/2026-09-30/102489915.html"},"quality_warning":null},"export":{"title":"大模型产业的开源与闭源之争 - opinion.caixin.com","url":"https://opinion.caixin.com/2026-09-30/102489915.html","summary":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","source":"opinion.caixin.com","date":"2026-09-30T03:55:00+00:00","content":"大模型自身正在形成一个价值数万亿美元的巨大产业，并且决定了整个AI价值链的价值锚点。是时候讨论大模型的产业格局了。\n2026年初以来，在Anthropic打开大模型编程的天花板，创造了可以被称为硅基劳动力的现象级产品之后，其年化收入（ARR）奇迹般地逐月跳升，很多人认为这种指数级的增长在几年内都不会放缓，也一度认为大模型的产业格局将会很快收敛到Anthropic和OpenAI的双寡头格局。\n但7月似乎是第二个“DeepSeek时刻”：Kimi K3、DeepSeek V4 Pro、智谱GLM 5.3、Qwen3.8、Muse Spark1.2相继发布，每一款开源模型均在代码任务和智能体任务上匹敌Claude Opus 4.7，而上一代Opus 4.6曾被认为在智能体能力上是跨越了奇点时刻的模型，现在看对开源模型也非难事。","confidence":0.9,"diagnostics_url":"/api/diagnose?url=https%3A//opinion.caixin.com/2026-09-30/102489915.html","quality_bucket":"high","failure_kind":"none","retryable":false,"quality_reason":"High confidence: full text extraction produced 365 characters.","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 365 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":365,"summary_length":365,"usable_text_length":365,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":365,"summary_length":365}},"tags":[],"format_contract_version":"news_item_formats.v1"}}}