Amazon Adds Kimi & GLM to Bedrock as Jassy Calls for Open AI Models

Hamza KhaliqHamza Khaliq
November 1, 2026
135 views
Amazon Adds Kimi & GLM to Bedrock as Jassy Calls for Open AI Models
Verified Frontier Research • Cross-Verified Primary Telemetry • Zero Vendor Sponsorships
⚡

The 30-Second Executive Verdict & Direct Answer

Direct Answer: Amazon Web Services has officially expanded Amazon Bedrock with China's premier open-weights models, Moonshot AI's Kimi and Zhipu AI's GLM, while CEO Andy Jassy declared that Western enterprises are facing an acute shortage of high-performing Western open-source models. Jassy reaffirmed ambitions for Bedrock to eventually match AWS EC2's multi-billion-dollar scale.

• Core Breakthrough: Mainstream cloud hosting of premier Chinese frontier open weights directly within AWS sovereign compliance boundaries.

• Market Disparity: Jassy warned that while China aggressively open-sources 100B+ MoE models, Western labs remain heavily tilted toward closed proprietary APIs.

• Enterprise Footprint: Zero-setup API deployment on AWS Trainium2 and NVIDIA H100 clusters with native VPC encryption.

• Cloud Moat: Positioning Bedrock as the universal runtime broker across proprietary frontier giants (Claude, Nova) and international open weights.

Kimi + GLM
Models Added
EC2 Scale
Growth Target
100B+
MoE Weights
100% AWS
Sovereignty

SEATTLE — In a major strategic expansion of its cloud artificial intelligence infrastructure, Amazon Web Services has formally integrated Chinese open-weights foundation models Moonshot Kimi and Zhipu AI GLM into its flagship Bedrock platform. Disclosed during Amazon's third-quarter earnings disclosures, CEO Andy Jassy used the announcement to issue a rare geopolitical assessment of enterprise AI, warning that Western enterprises are "very hungry" for competitive Western open-source models as Chinese research labs increasingly dominate the global open-weights landscape.

The addition of Kimi and GLM addresses a growing structural dilemma for global Fortune 500 engineering teams: while proprietary models like Anthropic's Claude 3.5 Sonnet and OpenAI's GPT-4o provide stellar reasoning, enterprise procurement mandates increasingly require inspectable open weights that can be self-hosted, fine-tuned on confidential IP, and operated with zero telemetry bleed. In the absence of tier-1 open foundation releases from American frontier labs beyond Meta's Llama, corporate demand has surged for high-performing Asian open architectures.

Andy Jassy: The Geopolitical Open-Source Imbalance

Addressing analysts during the quarterly conference call, Jassy outlined AWS's long-term thesis that Bedrock will eventually rival Amazon's massive Elastic Compute Cloud (EC2) business. However, Jassy pointed directly to the stark divergence between Western and Eastern AI dissemination philosophies:

"Customers in the West are very hungry for more Western open source models," Jassy emphasized. "China has been very prolific in publishing open-weights models that benchmark exceptionally well, and enterprise developers want access to that diversity. Our goal with Bedrock is to be the undisputed destination where developers can run any frontier model—whether closed or open—with enterprise-grade security, VPC isolation, and data governance."

Andy Jassy at AWS Keynote Address
Amazon CEO Andy Jassy outlines AWS Bedrock multi-model strategy during Q3 corporate briefing • Source: Benzinga / Yahoo Finance

Technical Profile: Moonshot Kimi and Zhipu GLM on Bedrock

The newly available models represent China's highest-ranked open foundation systems, specifically engineered for ultra-long context comprehension, bilingual technical documentation, and mathematical reasoning:

Model / Provider Architecture Type Context Window License Terms Inference Hardware
Moonshot Kimi-1.5 Dense Transformer 2,000,000 Tokens Open Weights (Commercial) AWS Trainium2 / H100
Zhipu GLM-4-Plus MoE (Mixture of Experts) 128,000 Tokens Open Weights (Apache 2.0) AWS Trainium2 / H100
Meta Llama 3.3 70B Dense Auto-Regressive 128,000 Tokens Llama Community License AWS Inferentia2 / A10G
Anthropic Claude 3.5 Sonnet Proprietary Hybrid 200,000 Tokens Closed Cloud API Managed AWS Infrastructure

The Enterprise Cloud Landscape

Amazon's aggressive diversification of Bedrock highlights a critical divergence from Microsoft Azure's heavy reliance on OpenAI. By hosting Claude 3.5, Mistral Large, Meta Llama, and now premier Chinese architectures like Kimi and GLM under a single standardized orchestration API, AWS is cementing itself as the neutral infrastructure layer for corporate artificial intelligence.

Verified Primary Source Documentation

Synthesized from official Amazon Q3 2026 earnings disclosures and AWS Bedrock service releases.

Open Source Documentation →
⚡ AI Central Intelligence

Explore AI Models & Autonomous Tools

Benchmark, compare, and discover verified tools powering the new wave of frontier reasoning and autonomous software engineering.

Hamza Khaliq

AUTHORED BY

Hamza Khaliq

Student, Author, Learner, Developer and Researcher

Covering the latest AI developments and industry news.

More news you might like

OpenAI Rates GPT-6 Cyber Critical After 948 Linux Kernel Exploits
AI Agents & Autonomous CodingOpen-Source AI & Hardware
Nov 8, 2026•8 min read

OpenAI Rates GPT-6 Cyber Critical After 948 Linux Kernel Exploits

OpenAI has unveiled GPT-6 Cyber, an offensive security model rated Critical under its Preparedness Framework that successfully automated exploit chains in Firefox and the Linux kernel.

Mistral Large 4 Le Chonk Unleashed: 1T MoE Open Weights & Benchmarks
AI Agents & Autonomous CodingOpen-Source AI & Hardware
Nov 8, 2026•9 min read

Mistral Large 4 Le Chonk Unleashed: 1T MoE Open Weights & Benchmarks

Mistral AI releases Mistral Large 4 (Le Chonk), a 1-trillion-parameter mixture-of-experts model featuring 49 billion active parameters, open weights, and multimodal capabilities.

Anthropic AI Benchmark Velocity Hits 1.9X Pace via Internal R&D
AI Agents & Autonomous CodingOpen-Source AI & Hardware
Nov 8, 2026•8 min read

Anthropic AI Benchmark Velocity Hits 1.9X Pace via Internal R&D

Model Evaluation and Threat Research (METR) estimates Anthropic's AI benchmark improvement rate has surged to 1.9 times its historical pace, with 80% of acceleration fueled by internal model usage.

OpenAI Unveils GPT-6 Cyber: Autonomous Model Smashes ExploitGym
AI Agents & Autonomous CodingOpen-Source AI & Hardware
Nov 8, 2026•8 min read

OpenAI Unveils GPT-6 Cyber: Autonomous Model Smashes ExploitGym

OpenAI debuts GPT-6 Cyber, a specialized frontier security model that achieves a 51% benchmark leap and uncovers thousands of vulnerabilities in Linux and Firefox.