News listxAI releases Grok 4.3: Featuring forced reasoning and a 60% price cut to disrupt the market
動區 BlockTempo2026-05-03 02:03:04

xAI releases Grok 4.3: Featuring forced reasoning and a 60% price cut to disrupt the market

ORIGINALxAI 發布 Grok 4.3:主打強制推理、砍價 60% 殺入市場
AI Impact AnalysisGrok analyzing...
📄Full Article· Automatically extracted by trafilaturaGemini 翻譯1017 words
xAI has released Grok 4.3, featuring a 40% reduction in input token costs and a 60% reduction in output costs compared to the previous generation, alongside the launch of the Custom Voices suite, which can clone voices in 120 seconds. (Previous coverage: Musk's "World's Most Powerful AI" Grok 3 is now free for users: Until the servers crash! (Includes tutorial)) (Background: xAI co-founder departs! Musk publicly apologizes: The company had issues early on, many talents didn't even get an interview) Thinking every time and never skipping reasoning steps is the core product logic of Grok 4.3. xAI calls this design "Always-on reasoning," meaning that regardless of how simple a question is, the model will force a reasoning process before providing an answer, and this cannot be disabled. The knowledge cutoff is December 2025, but built-in web search and X search can supplement real-time information. — Elon Musk (@elonmusk) May 1, 2026 The API pricing for Grok 4.3 is $1.25 for input and $2.50 for output (per million tokens, under 200k tokens), totaling $3.75. Compared to the previous generation Grok 4.2's $2 input and $6 output, input costs have dropped by 37.5% and output costs by 58%, a significant overall reduction among mainstream models. Abacus AI CEO Bindu Reddy described it on X: "As smart as Sonnet 4.6, but 5x faster and 5x cheaper." Grok 4.3 launched today 🚀 It is as smart as Sonnet 4.6 and 5x cheaper and faster Gemini models coming soon — Bindu Reddy (@bindureddy) May 1, 2026 The Custom Voices suite launched alongside Grok 4.3 has a core selling point of a low barrier to entry: by providing a 120-second voice sample, the system completes voice modeling in less than two minutes, covering pronunciation, accents, and tone patterns, rather than just cloning the timbre. The TTS service offers five built-in voices: Eve, Ara, Rex, Sal, and Leo, priced at $4.20 per million characters; speech-to-text (STT) streaming is $0.20 per hour, and batch processing is $0.10 per hour. The custom voice API endpoint is currently limited to Enterprise plans, with each team allowed to create up to 30 voice IDs, compliant with SOC 2 Type II, HIPAA, and GDPR regulations. Geographically, it is currently only available in the United States, with Illinois excluded due to biometric privacy laws.
Data Status✓ Full text extractedRead Original (動區 BlockTempo)
🔍Historical Similar Events· Keyword + Asset Matching6 items
💡 Currently matching via keywords + symbols (MVP) · Will be upgraded to embedding semantic search later
Raw Information
ID:9f7f088362
Source:動區 BlockTempo
Published:2026-05-03 02:03:04
Category:zh_news · Export Category zh
Symbols:Unspecified
Community Votes:+0 /0 · ⭐ 0 Important · 💬 0 Comments
xAI releases Grok 4.3: Featuring forced reasoning and a 60% price cut to disrupt the market | Feel.Trading