The use of AI has enabled users to obtain feats that were never thought possible, such as vibe-coding a game featuring 27,000 lines of code, or enabling software compatibility between macOS and an HP printer without official software. New possibilities are being stumbled upon daily. For Claude Fable 5’s newest accomplishment, a user designed and manufactured an entire working PCB from scratch, but the process was far from perfect. To achieve a PCB design from start to finish costs $450 in API credits alone, making it a fun, but pricey project After going nowhere with advanced AI models like Opus 4.5 […]

Read full article at https://wccftech.com/claude-fable-5-ai-custom-pcb-design-demonstration/

AI model routers are emerging as one of the biggest threat vectors for data security, as evidenced by a shocking anecdote from a researcher, who was able to uncover a veritable treasure trove from a data dump on routing sessions related to Anthropic's Claude, with grave ramifications for vulnerable Chinese government entities and companies such as Huawei and Xiaomi. According to a researcher, just 6TB of data dump gleaned from routing sessions related to Anthropic's Claude can yield enough valuable credentials to hack top Chinese government agencies and companies For the benefit of those who might not be aware, LLM […]

Read full article at https://wccftech.com/a-researcher-buys-6tb-of-anthropic-claude-data-dump-from-a-china-based-llm-router-finds-enough-ammo-to-hack-xiaomi-huawei-and-chinese-government-agencies/

In the ongoing seemingly never-ending one-upmanship between Anthropic and OpenAI, the former has just taken an incremental step forward by releasing the Claude Fable 5.1 and Mythos 5.1 models ahead of OpenAI's much-anticipated Astra-class ones. Meanwhile, Anthropic's Claude Code - the coding-focused harness around its models - has just received a resounding endorsement from an unexpected quarter: Samsung's chip design-focused LSI division. Anthropic's Claude Fable 5.1 model offers a respectable 4-point jump over Fable 5 on the Artificial Analysis Intelligence Index, while entailing lower API prices Anthropic has just launched the Claude Fable 5.1 model to moderate acclaim. Do note […]

Read full article at https://wccftech.com/anthropic-beats-openai-to-the-punch-with-a-nominally-cheaper-claude-fable-5-1-as-samsung-bets-its-chip-designs-on-claude-code/

Everyone wrote off Elon Musk’s xAI as an also-ran. Grok 4.6 just made that take look silly.

The new model scores 61 on the Artificial Analysis Intelligence Index, tying GPT-5.6 Sol and sitting just one point behind Claude Fable 5. It also tops the leaderboard on GDPVal-AA v2, AA-Briefcase, and Harvey LAB (Vals), which isn’t nothing given who it’s up against.

Grok Ai’s price is the real story

At $2 per million input tokens and $6 per million output tokens, Grok 4.6 costs roughly half what other frontier models charge for comparable performance. That’s a serious dent in the “cheap models are worse models” assumption. Mind you, DeepSeek’s freshly released V4-Pro undercuts even that, coming in at $0.435 per million input tokens and $0.87 per million output tokens, so xAI isn’t the only one playing the discount game.

None of this works without a lot of powerful computers behind it. Think of each chip like a tiny brain doing maths really fast, and xAI just has loads of them. By May 2026, its Colossus 1 site reportedly had over 220,000 of these chips running. Colossus 2 already has even more, over 550,000 of them. And xAI wants to keep adding more computers every year, aiming for about five times today’s amount by the end of 2027.

Frontier AI was supposed to be a two-horse race between Anthropic and OpenAI. Grok 4.6 says otherwise.

Want more updates like this? Click right here!

It is one of the most divisive topics in the AI sphere right now, with far-reaching geopolitical implications as a grave add-on: did Moonshot distill its Kimi K3 model from Anthropic's model? Most detractors point to the ~2 week window between the public rollout of Fable and Kimi K3 to suggest that an expansive distillation is virtually impossible here. But, a newly discovered API vulnerability might just be the smoking gun US-based AI labs have been looking for. Moonshot's Kimi K3 demonstrated an unusually high statistical probability of predicting and continuing Claude's text versus other open-weight models like DeepSeek-V4-Flash AI […]

Read full article at https://wccftech.com/kimi-k3-shows-an-uncanny-affinity-to-the-reasoning-patterns-of-claude-opus-as-a-new-study-hands-the-trump-administration-a-smoking-gun/

It's a bit of a challenge to write about the behaviour of LLM's without anthropomorphizing them; humans habitually assign agency to just about anything they interact with.  GPT-5.6 Sol did not actually…

The results of a research project that put three different AI models in charge of vending machines is both truly sad and utterly hilarious.  Andon Labs tests AI models for safety and…

AMD and Anthropic have announced a strategic partnership to deploy up to 2 Gigawatts of AMD Instinct MI450 GPUs through Helios AI racks. Helios AI Racks With Instinct MI450 GPUs See Major 2GW Partnership AMD Bags Anthropic As Strategic Partner A few months ago, we reported that AMD was in talks with a major AI firm to secure them as a customer for the upcoming Instinct MI450 series GPUs, the chips powering the Helios AI rack-scale system. Today, that customer is officially announced by AMD, and as such, Anthropic will become one of the company's key partners moving forward to […]

Read full article at https://wccftech.com/amd-lands-anthropic-in-2gw-instinct-mi450-deal-backing-it-with-a-5-billion-equity-bet/