-
DeepSeek V4 Pro 0813 vs GPT-5.6 Sol on DeepSWE: Cost, Coding, and Routing
Together AI ran 904 DeepSWE rollouts comparing DeepSeek V4 Pro 0813 and GPT-5.6 Sol. Sol wins pass@1 (72.7% vs 62.8%) at 35 times the cost per rollout ($8.37 vs $0.24), but a cascade that tries Pro first and escalates to Sol on failure solves 83.0% of tasks for $3.35 each.
# -
Major Frontier Model Providers Adopt Watermarking Tech to Comply with EU Regulation
As of August 2, 2026, the EU AI Act Article 50 requires AI systems to mark synthetic outputs in a machine-detectable manner. Major vendors are implementing statistical watermarking methods, which influence natural language generation without affecting performance.
# -
“Open weights are nowhere near a sufficient solution”: Dario Amodei fires back on AI power
In a public exchange on X, Anthropic CEO Dario Amodei told investor Gavin Baker that open weights "are nowhere near a sufficient solution" to AI's tendency to concentrate power, since they just shift concentration toward whoever has the most compute and chips. He pointed to Anthropic's backing of California's SB 53, which adds requirements for models trained above 10^26 FLOPs.
# -
What does an agentic SDLC actually look like?
Ryan welcomes Suneet Malhotra, Senior Manager of Test Engineering at Motorola Solutions, to chat about building end-to-end agentic SDLC pipelines using MCPs, using Cohen’s kappa to evaluate multiple LLMs-as-judges, and how you can improve requirements by shifting QA left through a specification…
# -
Executives put the spotlight on AI’s reliability issue
Only 35% of leaders say AI consistently delivers business outcomes, earns regulator confidence and is controllable, according to a report from HFS Research and TCS.
#