Bijan Bowen runs Grok 4.7 through hands-on coding and visual tasks, distinguishing advertised API pricingToken pricing is the rate an AI provider charges for processing input tokens, generating output tokens, or reading cached tokens. from the cost of long agentic runs. He compares extra-high reasoningReasoning effort is the amount of internal computational work an AI model applies before producing an answer or action. with a faster mode and notes that token consumption, context size and subscription limits affect real task costCost per completed task measures the total AI, tool, infrastructure, retry, and repair expense for each verified useful outcome..
The model builds a browser-based operating-system demo, a C++ skating game and several Blender, Godot and browser games. Some outputs are engaging and functional, while others remain basic foundations that need refinement rather than finished products.
A robotic-arm test fails to grasp and move an object reliably from a single camera view, despite responsive control. An era-changing city scene shows reasonable variation, but a luxury-watch websiteFront-end code generation uses AI to create or modify the user-interface code of websites and applications from instructions or examples. exhibits geometry and rendering errors. These examples expose a gap between code generationCode generation uses AI or another automated system to create source code from instructions, examples, schemas, or higher-level specifications. and spatial precision.
Bowen's overall assessment is positive but qualified: Grok 4.7 improves on earlier versions, yet no one-shot result proves general superiority. Long runs consume substantial usage, and visual quality varies by task.
Watch on YouTube
