closed-source

Miles Brundage observes that while the capability gap between open and closed AI models is months, the gap between models runnable on datacenter versus personal hardware is years, which matters for emergency and privacy use cases.

A live dashboard and statistical analysis shows open-source coding models are closing the gap with closed models at 1.5x the rate, with a 27B model already surpassing Claude Opus on decontaminated benchmarks. Tool-call reliability remains the main bottleneck.

The author accuses Anthropic of intentionally degrading the performance of their AI model Fable after an initial trial period, citing this as evidence that closed-source AI companies are predatory and that open-source alternatives will prevail.

An article discussing Alex Karp's critique of AI safety narratives, arguing that real safety for enterprises means control over data and models, and using Anthropic's relationship with Figma as an example of closed models capturing customer value.

A tweet highlights a podcast episode from The All-In Podcast that critiques closed source AI and predicts a surge in demand for open source AI.

The source code of Vohive's author has been closed; previously installed versions still work. Users can compile from forked repos or use reverse-engineered versions. The closure is not due to security concerns but because the author was hurt by certain individuals.

A tweet argues that closed frontier labs become more efficient and incentivized when open source labs exist, questioning whether that makes Anthropic anti-capitalist.

The author criticizes Dario Amodei, CEO of Anthropic, for his arguments against open source AI models, claiming Amodei misunderstands open source and that models can be run locally and inspected.

An opinion piece criticizing Anthropic for refusing to open-source its AI models while competitors like Google, OpenAI, xAI, and Meta have released open-source models, arguing that Anthropic's stance could lead to a dangerous monopoly in AI.

A discussion about the existence of trustworthy rankings comparing closed and open large language models, and whether models in the 70B–350B parameter range are worth the cost.

Analyzes the gap between open weights and closed source LLMs using the Artificial Analysis Intelligence Index and other benchmarks, finding that the gap is shrinking on some metrics but stable on others.

Fenng explained the positioning difference between Tencent's WeLM (closed-source large model) and Hunyuan (open-source large model), pointing out that the closed-source model will not disclose technical details or participate in evaluations, and suggested understanding it through product experience.

Elie Bakouch critiques Sakana AI's Fugu system as a closed-source orchestration layer over closed-source models, arguing it lacks transparency and true AI sovereignty, with technical limitations in routing and cost efficiency.

After firing Junyang Lin, Qwen has locked down its large models and is no longer releasing open source models, while other Chinese AI labs continue to open source their latest models. Rumors suggest the small model team is gone and Qwen 3.6/3.7 may be the last open source models.

Elena Rossini investigates claims that European microblogging platform W Social may have switched to closed source, highlighting discrepancies between its public image and actual practices, including migration of EU officials' accounts.

The article argues that without open-source LLM competition, closed-source companies like Anthropic will become arrogant and overcharge customers, as exemplified by a $200/month subscription.

Papers With Code announced a new feature supporting closed-source evaluations, using the dense Microsoft MAI Thinking 1 tech report as an example with a special 'closed' tag.

Commentary on the emergence of paid, subscription-based skill marketplaces for closed-source skills, noting how quickly this model has appeared after an era of free GitHub repos.

Vitalik Buterin argues against the narratives that bugs are inevitable and that AI bug-finding necessitates closed-source, stating that writing secure code has become harder but not impossible.

Article argues that AI-generated code and closed-source software are inherently less secure, and that LLMs like Anthropic’s Mythos will exacerbate vulnerabilities, making open-source projects the only trustworthy option.