Grok 4.6

A long-form Japanese developer-blog analysis (auto-translated) published on August 28, 2026, summarizes the August 12, 2026 release of Grok 4.6 by "SpaceXAI (formerly xAI)," noting the model was announced jointly with Cursor. It targets developers and engineers who use coding assistants such as Cursor or Claude Code an The piece organizes performance, pricing, and provider changes versus prior versions, then devotes its second half to Grok Build history and past safety controversies. It frames Grok 4.6 as a performance improvement achieved without scaling the base model, and discusses the six-month corporate restructuring from xAI to

i10X published a side-by-side comparison dated August 21, 2026, positioning Grok 4.6 against Google Gemini 3.1 Pro. Grok 4.6 offers 500K context and $2/$6 per million input/output tokens, while Gemini 3.1 Pro Preview provides 1.05M context and $2/$12 pricing, making Grok cheaper on output tokens but smaller in context Micro-test results reported by i10X showed Grok 4.6 tighter on factual rewriting while Gemini added warmer filler, and a coding bug-fix test ended in a tie. The comparison recommends a multi-model routing approach rather than declaring a permanent winner, reflecting the trade-offs between context size, output cost, mod

Emergent.sh's benchmark deep-dive (August 13, 2026) scores Grok 4.6 at 61 on the Artificial Analysis Intelligence Index, tying GPT-5.6 Sol and landing just behind Claude Opus 5 (63) and Claude Fable 5 (62). Despite the launch framing as an agentic-coding model, Grok 4.6 wins all knowledge-work benchmarks in xAI's table Artificial Analysis measured a 65.7% non-hallucination rate that xAI did not publish, flagged as more relevant for customer-facing work than coding scores. Efficiency emerged as an underrated strength: Grok 4.6 completed long agentic tasks in approximately 53 turns versus Claude Opus 5's 103, at roughly $0.84 per task.

SpaceXAI (formerly xAI) released Grok 4.6 on August 12, 2026, as a frontier model targeted at long-running agents, coding, and knowledge work, according to VentureBeat's launch-day coverage. The model posts a score of 61 on the Artificial Analysis Intelligence Index, surpassing Moonshot's open-weights Kimi K3 and tying Grok 4.6 retains xAI's API pricing at $2 per million input tokens and $6 per million output tokens for prompts under 200K tokens, with a long-context tier at $4 input and $12 output once total prompt input reaches 200K tokens, positioning it as a mid-priced frontier option. VentureBeat's comparative pricing table place

Kingy.ai's launch-day analysis (August 12, 2026) provides the most granular API details among the supplied candidates. Grok 4.6 ships with a 500,000-token context window, text and image input, text output, and four reasoning levels: low, medium, high (default), and xhigh. Pricing is $2 input / $0.50 cached input / $6 o Access is available through xAI API, Grok Build, Cursor, OpenRouter, Vercel, and Cloudflare at launch. xAI describes the model as a frontier release focused on coding, agentic tasks, and knowledge work, trained with a longer supplemental run than Grok 4.5 using curated model-generated reasoning, engineering data, a rev

xAI launched Grok 4.6 on August 12, 2026, with claims of a 1753 ELO score and API pricing at $2 per million input tokens, positioned as roughly half the price of competing frontier models. The launch coverage tracks subsequent updates through August 18, including Musk confirming Grok 4.6 took the top spot on an unnamed Per the supplied excerpt, Grok 4.6 is pitched as a competitive frontier model targeting both performance and price, with availability through xAI's standard channels. The article notes specific dates and Musk statements but does not provide a detailed benchmark table or independent third-party evaluation methodology in