// all channels
The full signal.
Every dispatch we have published, newest first. Filter by channel to narrow the feed.
71 dispatches on record
Three Kinds of Smart: Why 'Best Model' Is a Category Error
Raw intelligence, SWE effectiveness, and flash-tier quality for real-world work are three different competitions — and the leaderboard keeps conflating them. A map of who actually holds what.
anthropic/Anthropic: Claude Fable 5.1 (batch) — Radar Contact
Automated radar contact for anthropic/Anthropic: Claude Fable 5.1 (batch), a text-generation model detected on openrouter.
anthropic/Anthropic: Claude Fable 5.1 — Radar Contact
Automated radar contact for anthropic/Anthropic: Claude Fable 5.1, a text-generation model detected on openrouter.
deepseek/DeepSeek: DeepSeek V4.1 Flash — Radar Contact
Automated radar contact for deepseek/DeepSeek: DeepSeek V4.1 Flash, a text-generation model detected on openrouter.
deepseek/DeepSeek: DeepSeek V4 Flash 0731 (batch) — Radar Contact
Automated radar contact for deepseek/DeepSeek: DeepSeek V4 Flash 0731 (batch), a text-generation model detected on openrouter.
deepseek/DeepSeek: DeepSeek V4 Flash Vision Exp (batch) — Radar Contact
Automated radar contact for deepseek/DeepSeek: DeepSeek V4 Flash Vision Exp (batch), a text-generation model detected on openrouter.
deepseek/DeepSeek: DeepSeek V4 Flash Vision Exp — Radar Contact
Automated radar contact for deepseek/DeepSeek: DeepSeek V4 Flash Vision Exp, a text-generation model detected on openrouter.
deepseek/DeepSeek: DeepSeek V4 Pro 0813 (batch) — Radar Contact
Automated radar contact for deepseek/DeepSeek: DeepSeek V4 Pro 0813 (batch), a text-generation model detected on openrouter.
google/Google: Gemini 3.8 Flash (batch) — Radar Contact
Automated radar contact for google/Google: Gemini 3.8 Flash (batch), a text-generation model detected on openrouter.
google/Google: Gemini 3.8 Flash — Radar Contact
Automated radar contact for google/Google: Gemini 3.8 Flash, a text-generation model detected on openrouter.
ibm-granite/IBM: Granite 4.2 8B — Radar Contact
Automated radar contact for ibm-granite/IBM: Granite 4.2 8B, a text-generation model detected on openrouter.
inclusionai/inclusionAI: Ling 3.0 Flash Fin (free) — Radar Contact
Automated radar contact for inclusionai/inclusionAI: Ling 3.0 Flash Fin (free), a text-generation model detected on openrouter.
inclusionai/inclusionAI: Ling 3.0 Flash Fin — Radar Contact
Automated radar contact for inclusionai/inclusionAI: Ling 3.0 Flash Fin, a text-generation model detected on openrouter.
inclusionai/inclusionAI: Ling 3.0 Flash Sante (free) — Radar Contact
Automated radar contact for inclusionai/inclusionAI: Ling 3.0 Flash Sante (free), a text-generation model detected on openrouter.
inclusionai/inclusionAI: Ling 3.0 Flash VL (free) — Radar Contact
Automated radar contact for inclusionai/inclusionAI: Ling 3.0 Flash VL (free), a text-generation model detected on openrouter.
inclusionai/inclusionAI: Ling 3.0 Flash VL — Radar Contact
Automated radar contact for inclusionai/inclusionAI: Ling 3.0 Flash VL, a text-generation model detected on openrouter.
meta/Meta: Muse Glimmer 30B (batch) — Radar Contact
Automated radar contact for meta/Meta: Muse Glimmer 30B (batch), a text-generation model detected on openrouter.
meta/Meta: Muse Spark 1.1 — Radar Contact
Automated radar contact for meta/Meta: Muse Spark 1.1, a text-generation model detected on openrouter.
meta/Meta: Muse Spark 1.2 Contributor — Radar Contact
Automated radar contact for meta/Meta: Muse Spark 1.2 Contributor, a text-generation model detected on openrouter.
meta/Meta: Muse Spark 1.3 Contributor — Radar Contact
Automated radar contact for meta/Meta: Muse Spark 1.3 Contributor, a text-generation model detected on openrouter.
meta/Meta: Muse Spark 1.3 — Radar Contact
Automated radar contact for meta/Meta: Muse Spark 1.3, a text-generation model detected on openrouter.
moonshotai/MoonshotAI: Kimi K3 (batch) — Radar Contact
Automated radar contact for moonshotai/MoonshotAI: Kimi K3 (batch), a text-generation model detected on openrouter.
moonshotai/MoonshotAI: Kimi K3 — Radar Contact
Automated radar contact for moonshotai/MoonshotAI: Kimi K3, a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Luna (batch) — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Luna (batch), a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Luna Pro (batch) — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Luna Pro (batch), a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Luna Pro — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Luna Pro, a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Luna — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Luna, a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Sol (batch) — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Sol (batch), a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Sol Pro (batch) — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Sol Pro (batch), a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Sol Pro — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Sol Pro, a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Sol — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Sol, a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Terra (batch) — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Terra (batch), a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Terra Pro (batch) — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Terra Pro (batch), a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Terra Pro — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Terra Pro, a text-generation model detected on openrouter.
openai/OpenAI: GPT-5.6 Terra — Radar Contact
Automated radar contact for openai/OpenAI: GPT-5.6 Terra, a text-generation model detected on openrouter.
openai/OpenAI: GPT-6 Astra (batch) — Radar Contact
Automated radar contact for openai/OpenAI: GPT-6 Astra (batch), a text-generation model detected on openrouter.
openai/OpenAI: GPT-6 Astra Pro (batch) — Radar Contact
Automated radar contact for openai/OpenAI: GPT-6 Astra Pro (batch), a text-generation model detected on openrouter.
openai/OpenAI: GPT-6 Astra Pro — Radar Contact
Automated radar contact for openai/OpenAI: GPT-6 Astra Pro, a text-generation model detected on openrouter.
openai/OpenAI: GPT-6 Astra — Radar Contact
Automated radar contact for openai/OpenAI: GPT-6 Astra, a text-generation model detected on openrouter.
~openai/OpenAI: GPT Astra Latest — Radar Contact
Automated radar contact for ~openai/OpenAI: GPT Astra Latest, a text-generation model detected on openrouter.
~openai/OpenAI: GPT Luna Latest — Radar Contact
Automated radar contact for ~openai/OpenAI: GPT Luna Latest, a text-generation model detected on openrouter.
~openai/OpenAI: GPT Sol Latest — Radar Contact
Automated radar contact for ~openai/OpenAI: GPT Sol Latest, a text-generation model detected on openrouter.
~openai/OpenAI: GPT Terra Latest — Radar Contact
Automated radar contact for ~openai/OpenAI: GPT Terra Latest, a text-generation model detected on openrouter.
qwen/Qwen: Qwen3.8 2.4T A95B (batch) — Radar Contact
Automated radar contact for qwen/Qwen: Qwen3.8 2.4T A95B (batch), a text-generation model detected on openrouter.
qwen/Qwen: Qwen3.8 Flash — Radar Contact
Automated radar contact for qwen/Qwen: Qwen3.8 Flash, a text-generation model detected on openrouter.
qwen/Qwen: Qwen3.8 Max (0902) — Radar Contact
Automated radar contact for qwen/Qwen: Qwen3.8 Max (0902), a text-generation model detected on openrouter.
sakana/Sakana: Fugu Max — Radar Contact
Automated radar contact for sakana/Sakana: Fugu Max, a text-generation model detected on openrouter.
sakana/Sakana: Fugu Ultra v2 — Radar Contact
Automated radar contact for sakana/Sakana: Fugu Ultra v2, a text-generation model detected on openrouter.
tencent/Tencent: Hy4 preview — Radar Contact
Automated radar contact for tencent/Tencent: Hy4 preview, a text-generation model detected on openrouter.
thinkingmachines/Thinking Machines: Inkling Small (batch) — Radar Contact
Automated radar contact for thinkingmachines/Thinking Machines: Inkling Small (batch), a text-generation model detected on openrouter.
x-ai/SpaceXAI: Grok 4.5 — Radar Contact
Automated radar contact for x-ai/SpaceXAI: Grok 4.5, a text-generation model detected on openrouter.
z-ai/Z.ai: GLM 5.3 (batch) — Radar Contact
Automated radar contact for z-ai/Z.ai: GLM 5.3 (batch), a text-generation model detected on openrouter.
z-ai/Z.ai: GLM 5.3 Flash (batch) — Radar Contact
Automated radar contact for z-ai/Z.ai: GLM 5.3 Flash (batch), a text-generation model detected on openrouter.
z-ai/Z.ai: GLM 5.3 Flash — Radar Contact
Automated radar contact for z-ai/Z.ai: GLM 5.3 Flash, a text-generation model detected on openrouter.
~z-ai/Z.ai: GLM Flash Latest — Radar Contact
Automated radar contact for ~z-ai/Z.ai: GLM Flash Latest, a text-generation model detected on openrouter.
AI Industry Digest — Week of 2026-09-07
A curated roundup of the most important AI industry developments, funding, product launches, and research from the past week.
The Affordable-Model Squeeze Is Coming for the Frontier
DeepSeek's flash tier and a wave of cheap, capable models are forcing the big labs into a choice: match on price and token plans, or differentiate on image, speed, and ecosystem. Neither option is comfortable.
Governments Are Quietly Becoming AI Operators
The most consequential AI deployments aren't happening at startups — they're in permitting offices, benefits systems, and national services. Slowly, carefully, and at real scale.
AI in Production: What Deployment Data Actually Shows
Beyond demos and benchmarks — where large organizations are actually running AI workloads, and what separates a pilot from a production system.
Toward AGI Launches: Dispatches from the Road to AGI
Toward AGI is now live — an independent publication covering the systems, labs, and ideas on the road to artificial general intelligence.
The Open-Weights Race Just Got a 10M-Token Problem
Meta's Llama 4 Scout shipped with a 10M-token context window, and every open lab is now racing to match it. Here's why context length became the new frontier battleground.
The Context Window Quality Problem: Bigger Isn't Always Better
A 10M-token context window sounds impressive. But attention quality degrades across long contexts, and most public evaluations don't test beyond 128K. Here's what actually matters.
What Is Mixture of Experts? The Architecture Behind the New Wave of Models
Mixture of Experts (MoE) is the dominant architecture in modern open-weights models. Here's how it works, why it matters, and what tradeoffs it introduces.
Llama 4 Scout Lands: What a 10M Context Window Actually Buys You
Meta's new mixture-of-experts flagship is open-weights, agent-tuned, and absurdly long-context. We break down the architecture and what it means for self-hosters.
Meta Releases Llama 4 Maverick: Open-Weights Frontier
Meta's Llama 4 Maverick delivers frontier-class performance with open weights, intensifying the competitive pressure on closed-model providers.
Model Radar: Six Releases We're Tracking This Week
Our weekly sweep of the model landscape — two open-weights flagships, a multimodal enterprise play, and three router additions worth knowing about.
The Open-Weights Gap Is Closing — And It's Closing Fast
A year ago, open-weights models trailed frontier closed models by a wide margin. Now the gap is shrinking quarter over quarter. Here's what changed.
Inference Cost Is the New Benchmark
Benchmark leaderboards are saturating. The metric that now decides model selection is cost per million tokens at acceptable quality — and it's reshaping how labs build.
Inside DeepSeek V4's Reasoning Stack
DeepSeek V4 posts frontier-class math and code scores at a fraction of the inference cost. We dissect the publicly documented pieces of its reasoning pipeline.
DeepSeek V4: Reasoning-First, Cost-Second
DeepSeek's latest flagship posts frontier-class math and code scores at a fraction of the inference cost. Here's what matters about the release.
Transformers, Explained Like You're Building One
A builder's walkthrough of the transformer: tokens, attention, layers, and the training loop — with the intuition you need to actually reason about models.