{"slug":"qwen-3-8","label":"Qwen 3.8","item_count":14,"day_count":10,"source_count":2,"first_seen":"2026-08-14T16:16:29+00:00","last_updated":"2026-09-02T11:03:39+00:00","generated_at":"2026-09-03T20:11:53.837436+00:00","sources":["search_cn_open_weight_labs","simon_willison"],"days":[{"date":"2026-08-14","items":[{"title":"Run Qwen 3.8 27B on AMD Ryzen™ AI Max Agentic PCs and Radeon ™ GPUs - AMD","url":"https://news.google.com/rss/articles/CBMiqwFBVV95cUxQYzQ5c1BQdlI0THVOcm5Ia2lfQzRMUUp3NjM4Uk5yWUxCdl9hNEh1RkFJVFBZVmEyZHZFcU9HdWloUVdCM3FnY3o5LXczd0dLYW04OEt1YURkdjlLREJ0R2g1X1oyaXBEelp1V0lIc0VmUXpwcmwxTDdmV0JmS0ctVDBiVTFfUmZJRUsxVFprOVBjeFBKYkkzaXZuQU9HMUV0Mkw4RVJHSnYwZkE?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"Run Qwen 3.8 27B on AMD Ryzen™ AI Max Agentic PCs and Radeon ™ GPUs AMD","why_it_matters":"Matches feed focus: agentic.","sid":"a9c8b35dcb9918d4","published":"2026-08-14T16:16:29+00:00","editor_note":"AMD ships a same-day guide for running Qwen 3.8 27B locally on Ryzen AI Max hardware."},{"title":"Alibaba's Qwen team releases Qwen 3.8 models with open weights under the Apache 2.0 license - the-decoder.com","url":"https://news.google.com/rss/articles/CBMitgFBVV95cUxPeVVWdThLeXZqbHlybUdPMkRCSXBheDh1LXNjU1EzT1JFMHZUX1ZfWGQ4ekNNRlFjd3pKbmZPbHFJd2Y4QlVzbTNjejBGZWFMU2YyelEtNlp5Y21JWENQNzFHSXVJcW85bk8wcFBxVEVCaG5TSEdxRGRHMGY1Vk53emRzeTFEXzFWVl9UQlBBZUFneGNEWUZnYzFpVWU5UG5DX2NkQ1ZVc1NBV19tYzdWdU5QZkR1Zw?oc=5","source":"search_cn_open_weight_labs","type":"release","summary_1line":"Alibaba's Qwen team releases Qwen 3.8 models with open weights under the Apache 2.0 license the-decoder.com","sid":"4f2d09b57d0c456e","published":"2026-08-14T17:03:47+00:00","editor_note":"Alibaba's Qwen team formally releases Qwen 3.8 with open weights under the Apache 2.0 license."}]},{"date":"2026-08-16","items":[{"title":"Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things","url":"https://simonwillison.net/2026/Aug/16/qwen-38-27b","source":"simon_willison","type":"news","summary_1line":"Friday's big release was Qwen 3.8 27B , an Apache 2 licensed 27B parameter vision-capable LLM from Alibaba's Qwen research lab. I've been looking forward to this one: 27B is an excellent size for running a model on a...","why_it_matters":"Matches feed focus: agent, codex.","sid":"45cd7975f6d2c641","published":"2026-08-16T22:00:39+00:00","editor_note":"Independent hands-on: Qwen 3.8 27B is strong but wastes tokens overthinking simple prompts -- the first real-world usability caveat."}]},{"date":"2026-08-17","items":[{"title":"Qwen 3.8 27B scores 52 on the Artificial Analysis Intelligence Index","url":"https://simonwillison.net/2026/Aug/17/qwen-38-27b-scores-52","source":"simon_willison","type":"news","summary_1line":"Qwen 3.8 27B scores 52 on the Artificial Analysis Intelligence Index That's the same score as GPT-5.6 Luna (max), and just one point behind GLM-5.2 (max) and DeepSeek V4 Pro 0813 (max) - that GLM is 753B and that Deep...","sid":"4fdeab7a66fc6733","published":"2026-08-17T23:58:14+00:00","editor_note":"Willison's follow-up benchmark post scores the 27B build at 52 on the Artificial Analysis Intelligence Index, one point behind much larger DeepSeek V4 Pro and GLM-5.2 models."}]},{"date":"2026-08-19","items":[{"title":"Claude Opus 5 vs GPT-5.6 Sol vs Qwen 3.8 Max: 4x Price Gap [2026] - tech-insider.org","url":"https://news.google.com/rss/articles/CBMihAFBVV95cUxOcmFyS0F6cE5MMmhBNTdEaFkxNzhFbnZQM3FXOXJCUTRmT1gxUzJvVW1RbEFObHZ3VkVXdjI2M1kwMHl1N2ZaM2VaTEdFMFNxY0pPem51MU1acGZ5dV9WS3VGUm5ZRkNRZkp6Qk1EbFZ3NExVc040MVFEa1JCcjA3NDIzYmw?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"Claude Opus 5 vs GPT-5.6 Sol vs Qwen 3.8 Max: 4x Price Gap [2026] tech-insider.org","sid":"229f9c3b91ade44e","published":"2026-08-19T00:22:54+00:00","editor_note":"First pricing angle: coverage quotes a 4x price gap between the flagship Qwen 3.8 Max and Claude Opus 5 / GPT-5.6 Sol."},{"title":"Qwen 3.8: How a 27B Open Model Rivals GPT-5.6 and Claude Opus - Intelligent Living","url":"https://news.google.com/rss/articles/CBMifEFVX3lxTE5ORXVMLUpicUJCbUItTlBJMVM1VHRCQjhYYkIyM0pRTzdQaF81QTZfRWVXNjBDSHFfanlHdUdWU0ZzS1dHeW9CN29lWHA3UU5HM0dQbFFsUFRqd3hTS3NqVnNlNnloemE5OW5MTlEwcExjN3lqUms1dzZ2dmg?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"Qwen 3.8: How a 27B Open Model Rivals GPT-5.6 and Claude Opus Intelligent Living","sid":"e6edd89770217e67","published":"2026-08-19T05:34:40+00:00","editor_note":"Coverage reframes the 27B build's benchmark parity as rivaling closed frontier models GPT-5.6 and Claude Opus, not just open-weight peers."}]},{"date":"2026-08-20","items":[{"title":"DeepSeek V4 Pro vs Qwen 3.8 Max: Pricing and Open-Weight Changes Shift the Comparison - Memeburn","url":"https://news.google.com/rss/articles/CBMiqgFBVV95cUxQelVKWEc1eWZ1NG9KNEtmVGhBWHZPR2RJTXlhMmNuSy16UjZoeVlXZ1dkbllCUF9Ja01DNmM3VzJfam92bVFCQl9aTFg5RUlsNUJUTEluVUFlSmJidDhlNVljSWpNMUVkU0FWbkhrX1RTUXZIUFFTemdGZU0wT0RScWF3bUdjODJ4WE8xbG11OWZVZUo0YVRGajBHNlVxMWQtcnBKY1RHR2ROZw?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"DeepSeek V4 Pro vs Qwen 3.8 Max: Pricing and Open-Weight Changes Shift the Comparison Memeburn","sid":"cd4fdbe9658bad1a","published":"2026-08-20T10:15:31+00:00","editor_note":"Memeburn pivots the comparison to DeepSeek's price-hiked V4 Pro, positioning Qwen 3.8 Max as the cheaper open-weight alternative."}]},{"date":"2026-08-22","items":[{"title":"Qwen 3.8 Closes the Reasoning Gap, but Agentic Coding Remains a US Stronghold - forkast.news","url":"https://news.google.com/rss/articles/CBMinwFBVV95cUxNZjh4T2ktbGhlTHhNRThtci1wRkEwZDR2Um5fdGcycDdQNGJCcDllWVRMd3F1M3JnLWZuRUxBOTFsR2c4cGVQZDRWWTdad3JIT1pNdENGOVJQNjZDOFdlRkV4czV1b041N3dNWlZZZUlPckVMTFZjTVhOdzhnNHlTR3Q0YTRFLXVBYlNGbnBYOWlwZXB0eERSUThkSGpYZzQ?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"Qwen 3.8 Closes the Reasoning Gap, but Agentic Coding Remains a US Stronghold forkast.news","why_it_matters":"Matches feed focus: agentic.","sid":"6bd7d060abf5f19f","published":"2026-08-22T05:46:00+00:00","editor_note":"forkast.news splits the frontier-parity claim by capability: reasoning gap closed, agentic coding remains a US strength."},{"title":"Qwen 3.8 Closes the Reasoning Gap, but Agentic Coding Remains a US Stronghold - CryptoRank","url":"https://news.google.com/rss/articles/CBMitAFBVV95cUxPbEctRUp6dkY3YTBnWDZmUTFqa2VsUjNQelBwWWsxQWdfNTJRNk4yMEp1V0lqRUFiWlFNS0dEeWllTUp5VUVCenNubmRsX0FfQXNmcWU0WXhaUnpiV0s2THZsREZCVEU1azEyLUZDRXN6ZGRzNFFQSG1mZEVuZXFCRXE3V3lBRzM1OWNuZzlSWW5vQThqNThnLVJtRjZyUDVLUkNzZXBESENRaW1TRmZKSnNCX1I?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"Qwen 3.8 Closes the Reasoning Gap, but Agentic Coding Remains a US Stronghold CryptoRank","why_it_matters":"Matches feed focus: agentic.","sid":"df365b66f6928f78","published":"2026-08-22T23:09:43+00:00","editor_note":"CryptoRank republishes the same forkast.news capability-split assessment; no new fact, but a second outlet is now carrying the claim."}]},{"date":"2026-08-26","items":[{"title":"Qwen 3.8 Flash-Next is Cheap, But There Are Complicating Factors","url":"https://news.google.com/rss/articles/CBMimAFBVV95cUxNa3Fzel8wMmpCU2t1dzdxVmZxcnVTVFkxWTRxSUNHOGZxU05ZN2JVRXdjR3R5eThDd0ZGSFVFZ0hjUXUtM09WNkJfMDJqQ0ZrcndIZGhEemtueHowYWhXbWF4RkNMbEJKd1k4ckxVNTh3SWhnZEJaS05meVRwZnpCZFJGTDhhcS11aF9VVlRERFZ6OVM5Q2lCcw?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"Qwen 3.8 Flash-Next is Cheap, But There Are Complicating Factors AI Business","sid":"c4fce5c84f5c9fe2","published":"2026-08-26T20:27:12+00:00","editor_note":"AI Business covers a cheaper Flash-Next variant, flagging unspecified complicating factors -- first crack in the pricing narrative."}]},{"date":"2026-08-27","items":[{"title":"Qwen 3.8 is ridiculous, and the reason I’m getting Claude Opus tier results from a gaming GPU","url":"https://news.google.com/rss/articles/CBMinwFBVV95cUxOdUV3OUV3RUY0RkIzejJRRFhPSnE1MTR4ZzYxT0w5OXhuelFqVi1XdXlaRlUzUUdjUFpzN2JZNjJzR1JOWlN2SnFJTW11eDhVSlQ2SUpHZjBzc3l3THIxejAyVTVqR21fcFN2N0dqV1NIRHhFY0QtdTBQajloaFhQOUJCaC1TSUs1Yy1qQjNiTFdxUDI4TDNNeGZDNmhRNnM?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"Qwen 3.8 is ridiculous, and the reason I’m getting Claude Opus tier results from a gaming GPU MakeUseOf","sid":"148d281878cc48b2","published":"2026-08-27T11:01:13+00:00","editor_note":"MakeUseOf hands-on claims Claude-Opus-tier results running Qwen 3.8 on a gaming GPU, no benchmark score cited."},{"title":"Breaking Claude Code Opus 5 Auto Mode","url":"https://simonwillison.net/2026/Aug/27/breaking-claude-code-opus-5-auto-mode","source":"simon_willison","type":"news","summary_1line":"Breaking Claude Code Opus 5 Auto Mode Anthropic are putting a great deal of faith in Claude Code's auto mode for protecting their coding agent users against prompt injection attacks. They recently made that the defaul...","why_it_matters":"Matches feed focus: agent, claude code.","sid":"86c9015dd55dff65","published":"2026-08-27T22:50:25+00:00","editor_note":"Willison's same-day post on Claude Code's auto-mode safety net -- included in this cluster but not itself a Qwen 3.8 development."}]},{"date":"2026-08-29","items":[{"title":"Qwen 3.8 Flash Reduces Costs to One-Third of DeepSeek-V4-Flash","url":"https://news.google.com/rss/articles/CBMimAFBVV95cUxQeE50b2tjajhZd0N6a0lvX3VuVU4xckJWMXRBTkJZOWotMVdSMXVaWThGOTVxMHlBNzA3NkxEekFwaTVyeW1GU3pERHZQa0pObHdSNTdVNHQ1V0s0NTEweDJ1UUFfdHJDWG5ER0xnV1dEM0J3c093RUZXUldpNzJSNXZxc0tRQkFNaGI4MlNLSkU3VXpXdHlfOA?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"Qwen 3.8 Flash Reduces Costs to One-Third of DeepSeek-V4-Flash KuCoin","sid":"bd75c16722c9eeec","published":"2026-08-29T05:15:10+00:00","editor_note":"First hard cost claim: Qwen 3.8 Flash reported at roughly a third of DeepSeek V4 Flash's running cost."}]},{"date":"2026-09-02","items":[{"title":"The Sequence Learning Loop - Issue 925: Learn About Fable and Mythos 5.1, GLM-5.3-Flash, and Qwen 3.8","url":"https://news.google.com/rss/articles/CBMie0FVX3lxTE15ODlkcDc5RXk4dzQwbUFXd2lveWlxaS15Z19xaEg2MEExcVZyUEtSaVhFTFh1WlV0TzlldWkxbmxtX0hSX0RiYS0ySW13SWtPWThMWGR4eWR4RkRZWkdyekVNWDZxc1pTelNlN1QzSFktX0FMSGZOcG9xUQ?oc=5","source":"search_cn_open_weight_labs","type":"news","summary_1line":"The Sequence Learning Loop - Issue 925: Learn About Fable and Mythos 5.1, GLM-5.3-Flash, and Qwen 3.8 thesequence.substack.com","sid":"9db6ac5df4033290","published":"2026-09-02T11:03:39+00:00","editor_note":"The Sequence's Sep 2 issue keeps Qwen 3.8 in its roundup alongside Fable/Mythos 5.1 and GLM-5.3-Flash -- no new claim, just continued coverage."}]}],"editorial":{"tldr":"Alibaba shipped Qwen 3.8's 27B build under Apache 2.0 in mid-August, and independent benchmarks quickly put it within a point of much larger DeepSeek V4 Pro and GLM-5.2. By August 22, coverage split the frontier-parity claim by capability: the reasoning gap had closed, but agentic coding still favored US models.","stale":false,"whats_new":"An Aug 29 report puts Qwen 3.8 Flash's running cost at roughly a third of DeepSeek V4 Flash, shifting the storyline from capability parity to a price fight at the budget tier.","why_it_matters":"Beyond August's reasoning-vs-coding split, Qwen 3.8 is now positioning on price at the budget tier -- relevant if you're weighing cheap open-weight options for high-volume, cost-sensitive agent calls.","take_for_builders":"Before swapping in Qwen 3.8 Flash for cost savings, verify the claimed 3x price edge over DeepSeek V4 Flash against official rate cards and check what complicating factors the Flash-Next variant carries before shipping it to production.","status":{"state":"Reasoning parity holds; new price edge over DeepSeek at the Flash tier","tone":"now","changed":"2026-08-29","detail":"Aug 22 coverage found reasoning parity with US frontier models but a lingering agentic-coding gap; Aug 26-29 coverage shifted to economics, with Flash-tier Qwen 3.8 priced at roughly a third of DeepSeek V4 Flash despite unconfirmed complications flagged the same week.","track":[{"label":"27B shipped, Apache 2.0","detail":"Aug 14","tone":"launch","weight":14},{"label":"hands-on: strong but overthinks","detail":"Aug 16","tone":"rising","weight":8},{"label":"benchmark parity confirmed","detail":"Aug 17","tone":"rising","weight":10},{"label":"frontier comparison","detail":"Aug 19","tone":"rising","weight":10},{"label":"priced vs DeepSeek V4 Pro","detail":"Aug 20","tone":"rising","weight":10},{"label":"reasoning closed, coding trails","detail":"Aug 22","tone":"turn","weight":12},{"label":"Flash-Next: cheap, complicated","detail":"Aug 26","tone":"rising","weight":8},{"label":"gaming-GPU hands-on vs Opus","detail":"Aug 27","tone":"rising","weight":8},{"label":"Flash undercuts DeepSeek 3x","detail":"Aug 29 -> now","tone":"now","weight":20}]},"beats":[{"kicker":"LAUNCH","tone":"launch","headline":"Alibaba formalizes the Qwen 3.8 27B release under Apache 2.0 as AMD ships a same-day local-run guide","sids":["a9c8b35dcb9918d4","4f2d09b57d0c456e"]},{"kicker":"HANDS-ON","tone":"rising","headline":"Independent hands-on: strong, but wastes tokens overthinking simple prompts","sids":["45cd7975f6d2c641"]},{"kicker":"BENCHMARK","tone":"rising","headline":"Artificial Analysis score lands one point behind DeepSeek V4 Pro and GLM-5.2 despite the size gap","sids":["4fdeab7a66fc6733"]},{"kicker":"FRONTIER COMPARISON","tone":"rising","headline":"Press coverage recasts Qwen 3.8 against closed frontier labs -- a 4x price gap on the Max variant, benchmark rivalry on the 27B build","sids":["229f9c3b91ade44e","e6edd89770217e67"]},{"kicker":"PRICED VS DEEPSEEK","tone":"rising","headline":"Comparison pivots again: Qwen 3.8 Max vs. a price-hiked DeepSeek V4 Pro","summary":"Memeburn's Aug 20 piece drops the closed-frontier framing and instead prices Qwen 3.8 Max against DeepSeek's newly hiked V4 Pro, positioning Qwen as the cheaper open-weight choice a day after V4 Pro's price hike was quantified at 14x its cheapest model.","sids":["cd4fdbe9658bad1a"]},{"kicker":"CAPABILITY SPLIT","tone":"turn","headline":"Coverage splits the frontier-parity claim: reasoning gap closed, agentic coding still trails","summary":"forkast.news is the first outlet to break the \"matches frontier\" claim down by capability area rather than a single score, crediting Qwen 3.8 with closing the reasoning gap while still rating agentic coding a US strength.","sids":["6bd7d060abf5f19f","df365b66f6928f78"]},{"kicker":"FLASH-NEXT","tone":"rising","headline":"AI Business: a cheaper Flash-Next variant ships, but the piece flags unspecified complicating factors","sids":["c4fce5c84f5c9fe2"]},{"kicker":"GAMING-GPU HANDS-ON","tone":"rising","headline":"MakeUseOf: Qwen 3.8 delivers Claude-Opus-tier results running on a consumer gaming GPU","summary":"A same-day Willison post on Claude Code's auto-mode safety net also surfaces in this cluster without being a Qwen 3.8 development.","sids":["148d281878cc48b2","86c9015dd55dff65"]},{"kicker":"FLASH COST","tone":"now","headline":"Qwen 3.8 Flash undercuts DeepSeek V4 Flash by roughly 3x on running cost","sids":["bd75c16722c9eeec"]},{"kicker":"NOW","tone":"neutral","headline":"The Sequence's Sep 2 roundup keeps Qwen 3.8 in the frame alongside Fable/Mythos 5.1 and GLM-5.3-Flash, without adding a new claim","sids":["9db6ac5df4033290"]}],"open_questions":["Does Qwen 3.8's Apache 2.0 license and the 27B build's overthinking default extend to the flagship 2.4T Max variant, or are they isolated to the smaller open-weight release?","Does the Artificial Analysis Intelligence Index include Kimi K3 for a direct score comparison, or does the head-to-head with Qwen 3.8's chief open-weight rival remain untested?","Are the recurring price-gap claims (4x vs Claude Opus 5 / GPT-5.6 Sol, vs. DeepSeek V4 Pro on Aug 20, vs. DeepSeek V4 Flash on Aug 29) based on official rate cards from all vendors, or third-party estimates?","Does forkast.news's agentic-coding gap rest on a cited benchmark (e.g. SWE-bench), or is it a qualitative assessment without a published score?","What are the 'complicating factors' AI Business flagged for the Flash-Next variant -- a licensing carve-out, capacity limits, or a quality regression?","Does the consumer-GPU 'Claude-Opus-tier' claim rest on a benchmark score, or is it a subjective hands-on impression?"],"generated_at":"2026-09-03T09:15:00+00:00"}}