#ai#llm#api#machinelearning
Tencent Hy4: Vision + Computer-Use API Today, $0.417/$1.251 Promo, 1M vs 65K Max-Out, Weights Pending
Attributed compile + one small offline run on my box Primary source: Tencent Launches Hy4, Joining...
Yong Yu7 min read
Author profile
Yong Yu
Developer articles and tutorials by yong_yu_f98e15562e9b120a0.
@yong_yu_f98e15562e9b120a0View on DEV
Articles by @yong_yu_f98e15562e9b120a0
Browse the latest writing surfaced through DevArt.
Yong Yu7 min read
#ai#llm#opensource#machinelearning
Hydrogen-1.0 MIT Open Weights: Gated BF16, NVFP4 262K Context Cliff, and Vendor FSR vs GPT-6 Astra
Attributed compile + one small offline run on my box Primary source: Introducing Hydrogen-1.0 by...
Yong Yu7 min read
#machinelearning#ai#llm#agents
LongCat-2.5 at $0.30/$1.20: Thinking Locks Temperature, 1M Context, and Vendor Benches Between Opus 4.8 and 5
Attributed compile + one small offline run on my box Primary source: Introducing LongCat-2.5 —...
Yong Yu8 min read
#ai#llm#opensource#machinelearning
Mistral Large 4 Open Weights: Vendor TB 41% vs AA's 25%, High Default Effort, and $0.70/$2.10 Before You Self-Host
Attributed compile + one small offline run on my box Primary source: Mistral Large 4 (Le Chonk) is...
Yong Yu7 min read
#ai#llm#google#cybersecurity
Gemini 4 Carbon Is #1 on AA Index — Still Fairwind-Only: Cyber Benches, $2/$10 $4/$20, and What to Do Before the API
Google DeepMind’s Gemini 4 Carbon (Oct 28) claims AA Index v5.0 #1 at 59 and SOTA cyber benches — still Fairwind/cyber-defenders first, no public model ID. Intro $2/$10 → $4/$20, 95% cache off, builder checklist before the API.
Yong Yu6 min read
#ai#llm#python#opensource
LFM3 at 2B: Pick the Branch, Skip Agentic Coding, and the 128K Hosted Cliff
Attributed compile + one small offline run on my box Primary (Liquid AI / Hugging Face, launch...
Yong Yu6 min read
#ai#llm#python#api
GPT-6.1 Terra at $1/$5: First Computer Call, the none Effort Knob, and the 272K Price Cliff
GPT-6.1 Terra is $1/$5 for computer use — half Sol. First Responses shapes (`type: computer` vs `computer_use`), Terra-only `none` effort, and an offline cost model for the 272K screenshot cliff. No live call.
Yong Yu6 min read
#ai#llm#python#api
Kimi K3.1 Pro From K2.x Code: Fields That Now Error, the Default max Effort, and Cache-Write Math for Agent Loops
Attributed compile + one small offline run on my box Primary source: Moonshot AI, Kimi K3.1 Pro...
Yong Yu6 min read
#ai#openai#llm#python
OpenAI Decisions API Is GA: First /v1/decisions Shape, Free Output Tokens, and When to Prefer It Over Responses
Attributed compile + one small offline smoke test Primary sources (OpenAI, 2026-10-22): Build more...
Yong Yu5 min read
#python#security#ai#devops
LiteLLM 1.107.0 Ran Without an Import: The .pth Startup Hook, a No-Python Audit, and a uv Cooldown
Attributed compile + three small runs on my box Primary source: Kunal Singh, Compromised LiteLLM...
Yong Yu7 min read
#ai#llm#programming#python
Medium Beat XHigh for Opus 5.5 on SlopCodeBench: Cost per Pass, the Noise Floor, and an Effort A/B for Your Repo
Attributed compile + two small offline scripts Primary source: Dex Horthy (with Gabe Orlanski),...
Yong Yu7 min read
#ai#llm#python#minimax
MiniMax M3.1 Flash Thinks at Max by Default: First Call, Effort as the Price Knob, and Cache Math vs DeepSeek Flash
Attributed compile + two small offline scripts Primary sources (MiniMax, 2026-10-19): MiniMax M3.1...
Yong Yu6 min read
#ai#llm#python#deepseek
DeepSeek V4.1-Flash-1017 Is Live Under the Same Name: First Call, Ignored temperature, and the Nov 1 Search Bill
Attributed compile + two small offline scripts Primary sources (DeepSeek, 2026-10-17):...
Yong Yu6 min read
#ai#llm#python#opensource
Muse Spark Open Is Out: Reasoning You Can't Turn Off, KV-Cache Math at 128K vs 1M, and Hosted vs Mac vs GPU
Attributed compile + two small offline scripts Primary sources (all 2026-10-15): Meta AI Research,...
Yong Yu6 min read
#ai#llm#python#voice
Grok Voice 3.0 Just Landed: First Realtime Session, Full Duplex, and Why $0.08/min Beats Guessing Latency
Attributed compile + one small offline smoke test Primary sources (xAI / SpaceXSI, ~2026-10-14/15):...
Yong Yu5 min read
#ai#llm#python#machinelearning
Coinbase's 9B Fraud Model Beat Opus 4.5: Copy the Pattern With a Verifiable Reward, Tier-Only Output, and Block-Only Harness
Attributed compile + one small offline run on synthetic data Primary sources (Coinbase Engineering,...
Yong Yu7 min read
#ai#llm#python#meta
Muse Wonder Just Landed: First muse-wonder-1.0 Call, Active Memory vs Active Focus, and Why max Isn't Wonder Full
Attributed compile + one small offline smoke test Primary sources (Meta, 2026-10-12): Muse Wonder: A...
Yong Yu6 min read
#ai#python#llm#microsoft
Microsoft-Decision-1 Just Landed: First systemone Call, Free Output Tokens, and the Qwen-to-MAI Rebase Catch
Attributed compile + one small offline smoke test Primary sources (Microsoft, 2026-10-09):...
Yong Yu6 min read
#ai#machinelearning#python#llm
Liquid AI's d1 Decision Models Went Open: Triage Support Tickets on a CPU With the 600M One (and Where It Fools You)
Attributed compile + one small real CPU run Primary sources (Liquid AI, 2026-10-07): Open d1: Edge...
Yong Yu6 min read
#ai#llm#python#claude
Claude Haiku 5.5 Just Landed: Migrating a Haiku 4.5 Classifier Without the 400s (temperature, Prefill, Thinking Blocks)
Attributed compile + one small offline smoke test Primary sources (Anthropic, 2026-10-07):...
Yong Yu6 min read
#ai#python#machinelearning#rag
EmbeddingGemma 2 Just Dropped: Local Code Search on a CPU, the Install Gotcha, and Two Ways It Fails Silently
Attributed compile + one small real run Primary sources (all Google, 2026-10-06): EmbeddingGemma 2...
Yong Yu6 min read
#gemini#ai#promptengineering#productivity
Gemini Skills Just Landed in Workspace: Your First SKILL.md, Moving a Gem, and Why a Migrated Gem Isn't a Working Skill
Attributed compile (not original research) Primary sources: Let skills in Gemini tackle your most...
Yong Yu7 min read
#openai#codex#ai#cli
Codex CLI's Big Refresh Just Landed: First Session, Parallel Agents, and Why /fork Isn't a Worktree
Attributed compile (not original research) Primary source: ChatGPT & Codex changelog by OpenAI,...
Yong Yu5 min read
#claude#anthropic#ai#codingagents
Claude Code Mods Just Landed: First Hook, Blast Radius Guard, and Why They're Not Sandboxed
Attributed compile (not original research) Primary source: Getting started with Claude Code mods by...
Yong Yu5 min read
#codingagents#gemini#google#ai
Gemini 4 Argon Just Dropped — But You Can't Call It Yet: Pricing, 1M Output Tokens, and What to Do Until the API Lands
Attributed compile (not original research) Primary source: Gemini 4 Argon: our next era of frontier...
Yong Yu5 min read
#claude#github#ai#codingagents
Claude Sonnet 5.5 Just Landed in Copilot: When to Use It, First Call, and Why List Price Lies
Attributed compile (not original research) Primary source: Introducing Claude Sonnet 5.5...
Yong Yu5 min read
#aws#ai#llm#agents
Grok 4.7 Just Landed on Amazon Bedrock: First Call, Reasoning Effort, and the Cost Trap
Attributed compile (not original research) Primary source: Grok 4.7 is now available on Amazon...
Yong Yu5 min read
#ai#openai#llm#python
GPT-6 Sol vs Grok 4.7: Which One Is Actually Cheaper? (Cache Pricing, Long-Context Cliffs, and Token Burn)
Attributed compile (not original research) Primary source: GPT-6 Sol vs Grok 4.7: Same Input Price,...
Yong Yu6 min read
#mcp#ai#python#llm
MCP Transports That Still Matter: stdio vs Streamable HTTP (and Why SSE Is a Trap)
Attributed Chinese → English compile (not original authorship) Original title: MCP Server...
Yong Yu5 min read
#ai#agents#python#llm
Make Agent Tool Calls Survive Production: Validation, Retry Taxonomy, and Side Effects
Attributed Chinese → English compile (not original authorship) Original title: AI Agent...
Yong Yu5 min readAdvertisement