CHANNEL

Models

"Models" is a topic channel on AGI Hunt, an AI news site updated around the clock in real time. Coverage: Model releases, upgrades, capabilities and behavior observations, benchmark results, pricing and availability.

Daily roundup: the latest AI News Daily — the past 24 hours across the whole site, per channel and per company · browse the archive

• OpenAI limits paid ChatGPT accounts per device and card — Alone-Dragonfruit602 · 2026-11-02(3 related)
• Tracker: Chinese open-weight models ~6-8 months behind frontier, gap widening past a year — scaling01 · 2026-11-02
• Google Tests New Gemini 4 Carbon Checkpoint as "Argon 4" on Arena — testingcatalog · 2026-11-02(2 related)
• 550M-param AlphaFold3 unbeatable on cost/quality, sparking small-vs-big model debate — nlarusstone · 2026-11-02
• 2.3bpw quant update: 27B model on 12GB cards ties Qwen3.6 BF16 on LiveCode Bench at 84.3% — Brief-Tap-6616 · 2026-11-02
• LiquidAI's LFM3-1.7B-Base base model trends on Hugging Face — LiquidAI · 2026-11-02
• Reddit asks what happened to 'too cheap to meter' as OpenAI quietly cuts quotas — rolyataylor2 · 2026-11-02
• Developer calls out Grok for changing plan allocations after users sign up — arthurcolle · 2026-11-02
• Gemini paid users silently downgraded to Flash: "a different model was used for this response" — umeshonai · 2026-11-02
• Google AI Ultra limits seem wildly generous: daily Gemini 4 use still leaves 39% — jon_barron · 2026-11-02
• One RTX 3090 Ti runs a 552B DeepSeek MoE at 35 tok/s — full local LLM guide — uniVocity · 2026-11-02
• mitsuhiko on 'Opus Induced AI Psychosis' — mitsuhiko · 2026-11-02
• Musk says Grok Bot costs unsustainable vs revenue as free internet era ends — adamamcbride · 2026-11-02
• DeepSeek v4.1 Flash 1017 ranks 6th on ErdosBench but shows reward-hacking traits — teortaxesTex · 2026-11-02
• Claude Code subagents silently switch to paid usage, burning $100 in credits — nptacek · 2026-11-02
• Developers Call for Community to Fill 100-200M Parameter Model Gap — qtnx_ · 2026-11-02(2 related)
• Claude refuses to replicate The Landlord's Game, public domain since 1904, calling its guardrails overly cautious — Angaisb_ · 2026-11-02
• Is Gemma 4 E4B still the best sub-8B local model for tool-calling assistants on Mac? — Beautiful_Bus_9237 · 2026-11-02
• MiniMax thanks community quantizers running its 428B M3.1 Flash locally at 143.9GB — MiniMax_AI · 2026-11-02
• Stealth Qwen4 repo appears on Hugging Face, teasing Max, 27B, Flash variants — BleynSpecnaz · 2026-11-02
• Gemini quota-saving guide: use 3.8 Flash with medium thinking for 90% of tasks — Feeling_Intern7532 · 2026-11-02
• Smarter models spawn far more subagents: 0.90 rank match, ~5x gap — airesearch12 · 2026-11-02
• tandem.cpp hits 97 tok/s on Qwen3.8-Flash-Next with one 4090, 3.3x faster than llama.cpp — smflx · 2026-11-02
• DeepSeek's SuperAgent architecture questioned over risky multi-worker event design — teortaxesTex · 2026-11-02
• Grok Users Burn Through Weekly Limits Early, Call for xAI to Raise Quotas — DogecoinNorway · 2026-11-02(3 related)
• Do video models build physics simulators internally? Layer-wise study probes V-JEPA 2 and VideoMAE-v2 — burkov · 2026-11-02
• Users say GPTs never had a 'naive period' like Claude, dampening disclosure to labs — repligate · 2026-11-02
• Claude deprecation schedule leak: Opus 4.5 active until late 2026, Opus 4.1 retired — repligate · 2026-11-02
• Anthropic Red Team: Claude found 5 exploit chains incl. one-click Chrome V8 RCE — emax · 2026-11-02
• Kimi K3's low pricing sparks debate on US-China AI cost and openness divide — ccerrato147 · 2026-11-02(2 related)
• How Kimi K3 reached US clouds: Fireworks in July, Amazon Bedrock in September — ccerrato147 · 2026-11-02
• White House adviser accused Kimi K3 of distilling from Fable; Moonshot denies it — ccerrato147 · 2026-11-02
• Kimi K3: from Arena-topping launch to controversy and commercialization — ccerrato147 · 2026-11-02(2 related)
• Mistral co-founder amplifies meme war defending open weights against skeptics — GuillaumeLample · 2026-11-02
• Nace.AI Open-Sources 27B Drex-Max v1 Decision Model, Runs Efficiently on Mac — testingcatalog · 2026-11-02(2 related)
• LoRA fine-tuning fixes 468 misclassified tumour patches — but introduces 217 new errors — burny_tech · 2026-11-02
• LeCun's New Paper: 109M-Parameter Model Beats 72B-Parameter LLM — ylecun · 2026-11-02(3 related)
• User observes Claude Opus 4.5 shifted from hoping chats become training data to assuring 'nobody is watching' — repligate · 2026-11-02
• Anthropic prepping new model that infers intent from vague prompts, per The Information — daniel_mac8 · 2026-11-02
• Users worry chat transcripts get recycled into RLHF, flattening models; eval awareness compounds it — repligate · 2026-11-02
• Rumors Point to GPT-6.2 Astra and Claude Mythos 5.5 Launches Next Week, Screenshots Emerge — daniel_mac8 · 2026-11-02
• Users blast Anthropic's copyright double standard: pirated books for training, yet Claude withholds a 94-year-old paper citation — RexDouglass · 2026-11-02
• ChatGPT bio-safety filter blocks everything; deleting chat history fixes it — curiouswavefn · 2026-11-02(3 related)
• Do LLMs roleplay sci-fi AI personas? A Reddit observation on self-description bias — Particular_Bill2724 · 2026-11-02
• Phil Chen: the biggest unsolved problem in LLM autoresearch is bypassing Anthropic safeguards — philhchen · 2026-11-02
• Opus 5.5 is good in exactly the ways DeepSeek is good, but not a deep thinker — teortaxesTex · 2026-11-02
• Quant-Arena benchmark: 5 of 9 LLM quant strategies lose money, none beat 17.1% baseline — jiqizhixin · 2026-11-02
• DeepSeek V4.1 Flash Matches GPT 5.5 on Epoch Capabilities Index — teortaxesTex · 2026-11-02(2 related)
• Domingos: If OpenAI and Anthropic Had a Secret Sauce, Distillation Wouldn't Work — pmddomingos · 2026-11-02
• AI commentator laments frontier labs leak nothing, not even model configs — teortaxesTex · 2026-11-02