drop-2026-09-15-kr-am
Dropkit drop
AI work notes by Hosang Kim
외부 자료를 모아 정리한 기록입니다. 직접 실행한 프로젝트 결과와 구분해 읽어 주세요.
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
Dropkit drop
[1] https://openai.com/index/jalapeno-first-results/
OpenAI published first results for Jalapeño, a Broadcom co-designed inference ASIC built for dense transformer inference at scale.
Dropkit drop
OpenAI presented its first custom inference chip at Hot Chips in late August 2026. The chip, named Jalapeño, is an ASIC co-designed with Broadcom and optimized for large-scale inf
Two Chinese labs shipped open-weight mixture-of-expert models at frontier scale in the same week, followed by a third. DeepSeek V4 Pro 0813 reached general availability with 1.6 tr
Alibaba shipped Qwen3.8-Max in early August 2026 as a sparse mixture-of-experts flagship with 2.4 trillion total parameters, 95 billion active per forward pass, and one million tok
A 2.4-trillion-parameter mixture-of-experts model from Alibaba reached API pricing of $2 input and $6 output per million tokens, with 95 billion active parameters and a 1-million-t
OpenAI published benchmark results for its Jalapeño inference chip, showing 1.5×–1.9× higher inference performance than Nvidia's GB300 and up to 3.6× better work-per-watt acr
Three mixture-of-expert architectures shipped from China inside two weeks, all with open weights and permissive licenses. OpenAI previewed throughput above 700 tokens per second on
DeepSeek shipped V4 Pro on August 13, 2026. The official release supersedes the preview and adds stronger agentic capability. The model is available via API, web interface, and mob
Qwen3.8-Max reached open weights on August 12–13, 2026. The architecture is a 2.4 trillion parameter sparse mixture-of-experts model that activates 95 billion parameters per forw
DeepSeek Harness, GLM-5.3, Qwen3.8-Max.
Two open-weight MoE models crossed 700B total parameters this cycle. One agentic benchmark crossed 80% on SWE-bench Verified. One Apache 2.0 open-weight model at 30B reached 76.0 o
AI firms still lack sufficient containment measures; no lab reached full implementation.
Gemini is the strongest primary-source-backed item here; most Claude, Grok, Tencent, Qwen, and GLM entries remain secondary and should stay downgraded where the label is only parti
Dropkit drop
Directly comparable “Gemini vs Claude vs ChatGPT vs Grok” single-source coverage was not fully confirmed in the gathered URLs, so the table keeps only source-backed claims and
Dropkit drop
Dropkit drop
Dropkit drop
The first week of August 2026 saw several updates across major AI model providers and platforms. There is a noticeable focus on coding and agentic workflows, with several models be
Gemini 3.7 Flash was released with API pricing and a focus on coding and agent workflows. The pricing was listed at $0.75 input and $3.75 output per 1M tokens. This model included
Claude Opus 5 launched on July 24, 2026. Pricing for Opus 5 remained unchanged. Performance results were reported to be strong for agentic coding and computer-use tasks.
DeepSeek V4 Pro API pricing moved to peak and off-peak tiers. Peak pricing is $1.32 per million input tokens and $3.96 per million output tokens. Off-peak pricing is $0.66 per mill
This note outlines recent observations regarding AI model API pricing, performance, and feature-specific costs. Data is drawn from public API documentation and reported figures.
Claude Opus 5 led the August 2026 Artificial Analysis Intelligence Index. This model carried a premium price. Its API was priced at $5 in per 1M tokens and $25 out per 1M tokens. T
Recent data points provide insights into the pricing and performance of prominent large language models (LLMs). This note synthesizes confirmed figures from various providers.
This briefing summarizes recent developments in AI models, focusing on pricing, performance, and key features as reported in the field. The information is drawn directly from the p
Introductory Flash rates vs Opus premium vs Grok 4.6. Public list prices, coding/agent positioning.
Grok 4.20 538 ms total / 513 ms TTFT; Gemini 2.5 Flash 667 ms. Claude Sonnet 5 and Grok 4.6 list prices.
This publication briefing summarizes observations from the field. It covers recent shifts in platform positioning and open-weight model statistics.
Gemini, Claude, Grok pricing and a one-page scoreboard from public sources.
2026-08-29 lab note: Kimi K3 / Qwen 2.4T MoE vs US open weights; Gemini, Claude, ChatGPT, Grok.
The strict same-week cross-vendor scoreboard is only partially verifiable from the gathered material; the strongest repeatable signals are pricing and agentic/coding, while search
Gemini, Grok, ChatGPT, Claude, and the new API/app items above are kept in one scoreboard lane, with robust items promoted and the rest downgraded to sensitive where the source tra
Dropkit drop
Scoreboard merge only: Gemini / Claude / ChatGPT / Grok plus this-week apps and APIs. Dual-pass same label is 🟢 robust; one-pass or flipped labels are ⚠️ sensitive.
Dropkit drop
Dropkit drop
Dropkit drop