#xAI

SpaceXAI 的 Grok Voice 3.0 是新的领先语音到语音模型，在 Artificial Analysis 语音到语音指数上以 86.9 首次亮相即排名第 1，同时在延迟方面也很有竞争力 Grok Voice 3.0 是 @SpaceXSI 的 Grok Voice 2.0 的继任者，这是一款可通过 SpaceXAI Realtime API 使用的语音到语音模型，价格为每分钟 $0.08。我们通过 SpaceXAI Realtime API 使用 Rex 语音对 Grok Voice 3.0 进行了评估。 关键要点： ➤ 语音到语音指数：Grok Voice 3.0 以 86.9 首次亮相即排名第 1，领先于 GPT-Live-1（Astra，medium）的 83.2、Gemini 3.8 Live Extended Thinking（High）的 82.6 和 GPT-Live-1（Sol，low）的 80.1。它高于 Grok Voice 2.0 的 81.3（+5.6 分）和 Grok Voice 1.0 的 73.7（+13.2 分），后者发布至今还不到六个月。该指数平均了语音推理（Big Bench Audio）、智能体性能（Tau Voice）、Arena 偏好和任务成功率 ➤ 语音智能体 Arena：Grok Voice 3.0 在偏好方面以 Elo 1070 排名第 3，落后于 Gemini 3.1 Flash Live Minimal（1100）和 Gemini 3.8 Live（1088），领先于 GPT-Live-1（Sol，low）的 1057，并以 94.8% 的任务成功率排名第 1，为榜单最高，略高于 Grok Voice 2.0 的 94.6%。Grok Voice 2.0 以 Elo 1011 落后 ➤ Tau Voice：Grok Voice 3.0 在我们实现的 Tau Voice 基准上以 71.6% 排名第 2，落后于 GPT-Live-1（Astra，medium）的 74.5%，领先于 Gemini 3.8 Live Extended Thinking（High）的 68.6% 和 GPT-Live-1（Sol，low）的 59.3%——高于 Grok Voice 2.0 的 56.5% ➤ Big Bench Audio：Grok Voice 3.0 在音频推理上得分 98.5%，领先于 Gemini 3.8 Live Extended Thinking（High）的 97.7% 和 Grok Voice 2.0 的 97.2% ➤ 速度：在 Big Bench Audio 上，Grok Voice 3.0 的平均首次音频时间为 0.81 秒，明显领先于 GPT-Live-1（Sol，low）的 1.24 秒、GPT-Live-1（Astra，medium）的 1.34 秒和 Gemini 3.8 Live Extended Thinking（High）的 1.35 秒，但慢于 Grok Voice 2.0 的 0.70 秒 ➤ 成本：Grok Voice 3.0 的输入音频成本为每小时 $4.80，与 Grok Voice 2.0 相同。这高于 GPT-Live-1（Sol，low）的 $4.47 和 Gemini 3.8 Live Extended Thinking（High）的 $3.50，低于 GPT-Live-1（Astra，medium）的 $5.83，并且比 GPT-Realtime-2.1 High 的 $10.75 便宜约 2.2 倍 祝贺 @elonmusk 和 @SpaceXSI 团队 🚀 更多详情见下方 ⬇️

由 @AIatMeta 推出的 Muse Wonder (Max) 已进入 Code Arena：WebDev 以 1651 分位列第 15，Text Arena 以 1513 分位列第 2！ 在 Code Arena：WebDev 中，Muse Wonder (Max) 以 1651 分位列第 15。这一结果略低于以 1659 分位列第 12 的 Muse Spark 1.3 (Max)，与以 1655 分位列第 14 的 Kimi K3 (Max) 相差 4 分以内，并比以 1640 分位列第 16 的 Grok 4.7 (xHigh) 高 11 分。 此次发布在 Text Arena 中相较过去的 Muse 模型有显著提升。与以 1494 分位列第 11 的 Muse Spark 1.3 (Max) 相比，Muse Wonder (Max) 提升了 +19 分并上升 9 位至第 2。它落后于以 1525 分排名第 1 的 Gemini 4 Argon (High) 12 分。 祝贺 @AIatMeta 团队发布这一版本！

SpaceXAI launched Grok Voice 3.0 tops Artificial Analysis’s Speech-to-Speech Index at 86.9, beating OpenAI and Google. The index blends spoken reasoning, simulated support calls, blind preference votes and task success, where GPT-Live-1 (Astra, Medium) reaches 83.2. The largest gain comes on τ-Voice, where a simulated caller asks the model to rebook flights or dispute charges using domain tools and a policy document. There, Grok’s success rate jumped from 56.5% under Grok Voice 2.0 to 71.6%. Grok’s edge is speed: it answers 40% sooner than GPT-Live-1, whose 1.34 seconds implies roughly 0.8 seconds, slightly slower than Grok Voice 2.0’s 0.70. Introducing Grok Voice 3.0 +2 It also costs $0.025 per task on Artificial Analysis’s Big Bench Audio run, against $0.030 for GPT-Live-1 and $0.055 for GPT-Realtime-2.1 (High).

Today we are announcing Harvey LAB-AA v1.1 in collaboration with Harvey. This updates our scoring methodology for the Legal Agent Benchmark (LAB) to add a hallucination check and require correct responses to not include material misstatements. LAB-AA v1.1's new headline metric, Hallucination-Gated All-Pass Rate, only credits a task when the deliverables satisfy every rubric criterion and contain no material hallucinations. We define a material hallucination as one that would mislead a reader on a substantive point, such as a wrong contractually required date, while a minor hallucination is a real error that is unlikely to meaningfully affect the legal interpretation of a deliverable. Minor hallucinations are reported separately and do not affect the headline score. Grok 4.7 (xhigh) leads at 9.4% Hallucination-Gated All-Pass Rate, and >60% of otherwise passing results across the models tested at launch contain a material hallucination. This is the first step in enhancing the methodology for Harvey LAB-AA. In future updates, we’re working with Harvey to better account for the full set of factors lawyers value, including usability features like style and tone. Key takeaways: ➤ Top of the leaderboard: Grok 4.7 (xhigh) from @SpaceXAI leads at 9.4% Hallucination-Gated All-Pass Rate, narrowly ahead of Muse Spark 1.3 (max) from @AIatMeta at 8.9% and GPT-6 Astra (max) from @OpenAI at 8.6% ➤ Hallucinations reshape the leaderboard: without the hallucination gate, Muse Spark 1.3 (max) would lead clearly with a 26.7% all-pass rate, but two thirds of those passes contain one or more material hallucinations, resulting in a Hallucination-Gated All-Pass Rate of 8.9%. GPT-6 Astra (max) retains almost all of its passes after the hallucination gate (8.9% to 8.6%) and moves from joint 10th to 3rd ➤ The GPT-6 model family is the most grounded: GPT-6 Astra (max) averages 0.03 material hallucinations per task (4 across all 120 tasks) and GPT-6 Sol (max) averages 0.07. When comparing six checker models on a 20-task subset, GPT-6 Astra had zero material hallucinations under every checker, including Claude Opus 5.5 (high). By comparison, Gemini 3.8 Flash (high) averages 13.96 material hallucinations per task. ➤ Top-scoring models aren’t the most expensive: Grok 4.7 (xhigh) takes the top spot on the leaderboard at a cost of ~$9.50 per task, under half the cost of Claude Fable 5.1 (max with fallback) at ~$21.70, the most expensive model. Muse Spark 1.3 (max) represents strong performance for its cost, landing in second for ~$4.20 per task Thank you to @nikogrupen and @ItsJulioPereyra from @harvey for their work on Harvey LAB and collaboration.

Grok 4.7 is behind only Anthropic models on AA-Briefcase, ranking just behind Opus 5 at ~50% of its Cost per Task Grok 4.7’s improvements over Grok 4.6 are clear in AA-Briefcase-Lite, our public due diligence scenario where models are tasked with building market models and target assessment decks. Grok 4.7 gains significantly in Analytical Quality Elo (1698 → 1994) with a slight regression in Presentation Elo (1531 → 1499). API cost to produce example decks: Grok 4.7 (xhigh) ~$8 vs. Grok 4.6 (xhigh) ~$4.40

Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.

Grok 4.7 by @SpaceXAI and @elonmusk is now in the Agent Arena! Your votes drive the @arena leaderboards, head over and bring your toughest prompts. In Agent Arena, we measure models on millions of real-world, long-horizon agentic tasks from a global community of users. Models can access web search, filesystem, and terminal tools to complete complex workflows. The leaderboard measures model performance on outcomes relative to the average model using a causal tracing methodology. In addition to Agent Arena, Grok 4.7 is in Battle Mode for: Text, Vision, Code, and Document.

Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.

Grok 4.7 is here. It's a notable improvement over Grok 4.6 at the same price and speed.

Grok 4.6 is live on Amazon Bedrock https://x.ai/news/grok-4-6-amazon-bedrock

Congrats to @SpaceXAI on one more amazing model: Grok 4.6. We benchmarked it as an orchestrator on our Wide-And-Deep-Research benchmark using the Perplexity Computer harness, and it neatly sits on the Pareto frontier of performance vs cost. Available to all Pro and Max users on Perplexity!

Grok 4.6 is now available in Perplexity and Perplexity Computer. On WANDR, it sits on the Pareto frontier of performance and efficiency, matching Fable 5 results at over 60% lower cost.

ICYMI refresh your model catalogs. Grok 4.6 is out 🛸 Built for long-running agent tasks, with stronger capabilities across coding, knowledge work, and visual tasks. $2/M input and $6/M output. Welcome Grok 4.6 to Pi.