The Daily Commit · Section Edition Front Page PHP AI Dev EN DE FR ES

TheModelDesk

August 12, 2026
models, agents & local inference

Releases

xAI Releases Grok 4.6 with Enhanced Agentic Capabilities

xAI has released Grok 4.6, an updated large language model emphasizing long-running agents and multi-step reasoning. The model matches GPT-5.6 Sol on composite intelligence benchmarks and shows particular strength in code generation, product development, and sustained interactive tasks. Available via Cursor, Grok Build, and API partners starting at $2 per million input tokens.

xAI released Grok 4.6 on August 12, 2026, advancing its capabilities from Grok 4.5 with emphasis on extended agentic workflows and visual development tasks.

The model underwent extended supplemental training using curated model-generated data for reasoning and technical concepts, alongside high-quality engineering data and refined optimization techniques. Supervised fine-tuning trajectories were regenerated using Grok 4.5 across reasoning, agent harnesses, and domains including STEM, software engineering, and knowledge work. The model was then trained on diverse agentic reinforcement learning tasks spanning knowledge work, general coding, kernel optimization, web development, and computer-aided design.

On composite benchmarking, Grok 4.6 matches GPT-5.6 Sol, achieving a 61 score on the Artificial Analysis Intelligence Index. The model demonstrates notable performance across specific agentic and coding benchmarks: 69.9% on CursorBench v3.2, 65.9% on DeepSWE v1.1, 61.3% on FrontierCode v1.1, and 57.5% on APEX-Agents. It shows particular strength in long-horizon project work, with improved first passes on visual and interactive applications and demonstrated self-testing behavior across extended task sequences.

Grok 4.6 is immediately available in Cursor and Grok Build, with 2× included usage credits for the first week. API access is available through the xAI console and via partners including OpenRouter, Vercel, and Cloudflare. Pricing is set at $2 per million input tokens and $6 per million output tokens, with a fast variant at double cost.

Safeguards were improved and calibrated to the model's expanded capabilities, with xAI conducting its widest pre-deployment testing suite and continued post-deployment evaluation.

Read the original source ↗

Rate this article: 0

Readers’ Forum

No contributions yet — open the debate.

← The Model Desk — Page C1

Models, agents & local inference · The Daily Commit · Screen edition · Imprint · Privacy Policy