Qwen4 Max
AI overview
Sign in and the AI will write an overview from our coverage.
Headlines · 5
- Test of 65 AI systems finds convergent advice on major choices
An author tested 65 AI systems by asking each one the same questions ten times without a system prompt. The author reports that 73 of 100 book recommendations were Man's Search for Meaning, 76 of 100 business ideas were subscription boxes, 90% of study recommendations were computer science, and 98% of investment answers favored an index fund. Answers showed less convergence for movies, dinners, baby names, testimonial names and brand colors. The report also found that harmless questions could reveal model-family similarities, with 59% of models most closely matching another model from the same lab versus 4% by chance.
Hacker News · AI(100+ 分) · 🔥 5 - DPO loss fell, but final-answer accuracy did not improve
The project author reports that DPO reduced validation loss from 0.628564 to 0.155683 and raised the validation preference score, but Qwen3-0.6B answered 29 of 32 controlled cases correctly versus 30 for the original model. The evaluation workflow included rule-based scoring, pairwise LLM judges, position-bias checks, human calibration and paired bootstrap; all six model comparisons were ultimately marked inconclusive. The 32-case study covered controlled chain-of-thought samples involving arithmetic and multistep numerical updates, and its human labels came from one annotator.
掘金 · 人工智能 · 🔥 0 - Qwen4 Max becomes available in OpenCode Go
Qwen’s official account relayed an OpenCode announcement that Qwen4 Max is now available through OpenCode Go. The announcement provides no pricing, performance results, access limits, or broader rollout details.
Qwen (X) · 🔥 0 - Alibaba releases Qwen4 Max with benchmark gains and lower task cost
Artificial Analysis reported that Qwen4 Max scored 43 on its Intelligence Index at xhigh reasoning effort, nine points above Qwen3.8 Max and three points below Kimi K3.1 Pro. It improved on every cited Index evaluation versus Qwen3.8 Max, used 39% fewer output tokens per task, and cost $3.27 per Intelligence Index task, about 11% less than Kimi K3.1 Pro. The model supports a 1M-token context window and text and image input with text output. Alibaba's first-party APIs list pricing of $1.50 per million input tokens and $4.50 per million output tokens.
ArtificialAnlys (X) · 🔥 0 - Qwen4 Max review finds gains over Qwen3.8 Max but long runtimes and high credit use
A hands-on review tested Qwen4 Max on visual recreation, game replication, multimodal controller interaction and larger game projects. The author found clear improvements over Qwen3.8 Max, including a more functional mechanical-watch recreation and a partially successful Super Mario recreation. The reviewer also reported problems with controller recognition and synchronization, while the larger game projects were limited or failed. Tasks often took about one to two hours. On a 59-yuan entry plan with roughly 2,000 credits available, the discounted run consumed 1,146 credits for a small number of examples, leading the author to question the model's cost-effectiveness.
掘金 · 人工智能 · 🔥 0
Experience and discussion from the community
Share my Qwen4 Max experienceAsk about Qwen4 Max
Nobody has shared their experience with Qwen4 Max yet.