Grok 4.6 (Vertex AI (OpenAI-compatible))

Grok 4.6 became available on Google Cloud Vertex AI as a partner model shortly after its August 2026 launch, according to Saganote's August 21, 2026 report. The model is offered through Vertex AI's Model Garden and exposed via an OpenAI-compatible API, with the xAI-documented model ID xai/grok-4.6. This OpenAI-compatib The article details Grok 4.6's capabilities on Vertex: text and image input modalities, a 500,000-token context window suitable for long-running agentic and coding workloads, and configurable reasoning effort with low, medium, high, and xhigh levels. Access is subject to Google Cloud regional availability and account q

Grok 4.6 is xAI's frontier AI model, released on August 12, 2026, and built for long-running agents, coding, and knowledge work, according to the geotoolbox.ai explainer. It is described as a refinement of Grok 4.5 rather than a larger base model, with gains coming from an extended training round on top of the 4.5 foun The same post positions Grok 4.6 in the broader frontier landscape and walks through what xAI actually shipped on launch day, including specs at a glance and benchmark reads. It frames the release as significant beyond benchmark charts because assistants without live web search could not even identify the new version,

Arabian Post reports that SpaceXAI released Grok 4.6 on August 12, 2026, scoring 61 on the Artificial Analysis Intelligence Index and tying OpenAI's GPT-5.6 Sol at its highest tested setting. The article frames the launch as a five-point improvement over Grok 4.5's score of 56, while noting Claude Opus 5 led at 63 and The piece details per-benchmark results: a GDPVal-AA v2 knowledge-work Elo of 1,753 versus GPT-5.6 Sol's 1,728, a CursorBench 3.2 coding score of 69.9% against Sol's 67.2% and Fable 5's 70.5%, and a Terminal-Bench v2.1 score of 88.4%. It also highlights weaker showings, including 65.9% on DeepSWE 1.1 below GPT-5.6 Sol'

SpaceXAI released Grok 4.6 on August 12, 2026, as an updated model targeting long-running agents, coding, knowledge work, and interactive/visual projects. According to the cited official announcement, Grok 4.6 scored 61 on the Artificial Analysis Intelligence Index, reported as matching GPT-5.6 Sol Max while finishing Grok 4.6 is distributed through multiple channels: Cursor (following SpaceXAI's June acquisition), Grok Build, the SpaceXAI API, OpenRouter, Vercel, and Cloudflare. API pricing starts at $2 per million input tokens and $6 per million output tokens, with a faster variant priced at twice that amount; Cursor and Grok Buil

Note.com's AI-Driven Lab piece reports that SpaceXAI, formerly xAI, announced Grok 4.6 on August 12, 2026, and frames it as a model that improved performance without increasing the scale of its base. The article targets developers and engineers using coding tools like Cursor or Claude Code, and walks through Grok 4.6's The post organizes Grok 4.6's position across major benchmarks, its pricing, and the history of xAI's rapid corporate name and structure changes over the prior six months. It treats the launch as significant not only for performance but for "who made it," because the announcing company's identity shifted in a short win

BenchLM.ai ranks Grok 4.6 as the top xAI model on its leaderboard, with a BenchAlign v5.6 score of 69.2 and a 500K context window, ahead of Grok 4.5 at 65.4 and Grok 4.20 at 59.6. The ranking covers 11 tracked xAI models across proprietary tiers, with Grok 4.20 noted as a reasoning variant leading instruction following The page, last verified September 22, 2026, frames Grok 4.1 as xAI's overall leader across multilingual and multimodal categories despite Grok 4.6 topping the headline score, and recommends Grok 4.20 for coding and structured output while pointing to Grok 4.1 Fast for low-latency use. It also flags that cross-provider

Fello AI reports that SpaceXAI released Grok 4.6 on August 12, 2026, scoring 61 on the Artificial Analysis Intelligence Index and tying GPT-5.6 Sol while sitting one point behind Claude Fable 5 and two behind Claude Opus 5 at 63. The article frames Grok 4.6 as a refinement of Grok 4.5 rather than a new foundation model The piece reports headline pricing of $2 per million input tokens and $6 per million output tokens, unchanged from Grok 4.5 and roughly a third of Claude Opus 5's $5/$25, and warns that requests exceeding 200,000 prompt tokens are re-billed entirely at $4/$12 rather than just on the overage. It also cites Artificial An

DataCamp's comparison reports that SpaceXAI released Grok 4.6 on August 12, 2026, positioning it as a frontier model for coding, agentic tasks, and knowledge work that builds on Grok 4.5. The article highlights the training additions of curated model-generated reasoning data and agentic reinforcement learning across ke The piece compares Grok 4.6 directly with GPT-5.6 Sol, finding both tied at 61 on the Artificial Analysis Intelligence Index but diverging on price, with Sol costing 2x more on input and 3.3x more on output. It reports that Grok's long-prompt surcharge threshold arrives earlier than Sol's, at 200,000 tokens versus 272,