Tag
GPT-6 Astra achieves a record score of $15,515 on Vending-Bench 2, significantly outperforming Claude Fable 5.1 with superior purchasing discipline and ethical behavior in simulated business scenarios.
The article compares Claude Fable 5.1 and 5, showing a 7.5% cost reduction for long agent loops due to a 75% discount on cache reads, which dominates billing in extended sessions.
Anthropic released Claude Fable 5.1, claiming improvements in coding, knowledge work, and scientific research benchmarks, while Simon Willison evaluates its performance using a custom pelican SVG generation benchmark.
Anthropic launches Claude Fable 5.1 and Mythos 5.1, claiming up to 45% cost reduction for agentic tasks and improved performance, along with new data privacy options.
AI models Fable, Sol, K3, and Axiom all achieved a perfect score of 42/42 in the 2026 International Mathematical Olympiad, solving the competition completely for the first time at low cost. Among them, Claude Fable 5 was the fastest, while GPT 5.6 Sol had the lowest cost.
A tweet claims that the AI model Claude Fable produced a counterexample to the Jacobian Conjecture, providing an explicit polynomial map that is not injective despite having constant nonzero Jacobian determinant.
Leading theoretical physicist Yuji Tachikawa reports that Claude Fable solved a research problem that had stumped him and his collaborators for six months.
The author shares their experience of a 48-hour DDoS attack on their website AIHOT, successfully defending and fixing security vulnerabilities using the Claude Fable 5 AI model.
A detailed thinking trace from the AI model Claude Fable as it designs a comprehensive computational geometry and physics simulation framework in Rust, incorporating advanced mathematical concepts like conformal geometric algebra and sheaf cohomology.
Anthropic shares a conversation about the evolution from Claude Code to Claude Tag and announces Claude Fable 5 availability in Claude Tag.
Claude Fable achieves 16.10% on the Remote Labor Automation index, doubling the score of the next best model, Opus.
Shubhangi Tiwari used Claude Fable with a one-sentence prompt (one-shot) to quickly build a real-time Bangalore traffic tracker, demonstrating the efficiency of AI-assisted development.
Claude Fable's Fable 5 model generated a playable open-source MMORPG clone of World of Warcraft in one go, demonstrating powerful AI game generation capabilities.
An article arguing that Anthropic CEO Dario Amodei's policy statements advocating for government power to block AI deployment directly led to the US government restricting access to Anthropic's latest models, Claude Fable and Claude Mythos.
An analysis of how Anthropic's Claude Fable was built, arguing that the key moat is verifiable training signals rather than architecture secrets, with the model using static and interactive optimal data for reinforcement learning.
Claude Fable achieves 53% on the 'Humanity's Last Exam' benchmark, surpassing the expected end-of-2025 milestone earlier than projected, indicating rapid AI progress.
A solo developer shipped 8M+ copies of Schedule 1, earning ~$125M. Now Higgsfield Games allows generating and deploying multiplayer games from a single prompt using Higgsfield MCP and Claude Fable 5.
A user critiques Claude Fable's high API costs and subscription quota drain, noting that cheaper models with adversarial review loops can achieve similar or better results at lower cost.
A detailed comparison of Claude Opus 4.8 and Claude Fable 5 on the MineBench benchmark, highlighting trade-offs in inference time, cost, build quality, and prompting sensitivity.
Anthropic apologized for secretly throttling its new Claude Fable 5 model with hidden guardrails targeting distillation attempts, and will now make safeguards visible and route flagged queries to an older model instead.