CHANNEL

Models

"Models" is a topic channel on AGI Hunt, an AI news site updated around the clock every 30 minutes. Coverage: Model releases, upgrades, capabilities and behavior observations, benchmark results, pricing and availability.

• Rumor says Codex voice, parallel agents, GPT-5.6 speedups and maybe Opus 5 arrive today — VraserX · 2026-07-23
• Claude Opus 5 appears to be rolling out across providers — kimmonismus · 2026-07-23
• Cross-model test says latest Claude refuses a prompt that many frontier models answer — cyb3rops · 2026-07-23
• GPT spends 20 minutes browsing the web before making a Kimi 3 vs. Fable 5 meme — D3VAUX · 2026-07-23
• Pichai says Gemini’s next leap needs much larger base models as Alphabet lifts 2026 capex to $205B — The Decoder · 2026-07-23
• Pichai admits Google still needs to improve at coding and agentic coding — firstadopter · 2026-07-23
• Four AIs are being scored against Polymarket, and so far they mostly agree with the market — soulsintention · 2026-07-23
• Rumors Claim Anthropic's Claude Opus 5 Launch is Imminent — kimmonismus · 2026-07-23(7 related)
• ByteDance's WorkflowGym Exposes GUI Agents: Top Models Pass Only 30% — 机器之心 · 2026-07-23
• Sam Altman said GPT-5.6 Sol would hit 750 tok/s in July, and July is almost over — daniel_mac8 · 2026-07-23
• GPT-6 leaks surface as Kimi K3 and Qwen 3.8 close in on GPT-5.6 — haider1 · 2026-07-23
• Reddit users say GPT-5.6 now blocks routine coding tasks far more often — Broad_Commission_242 · 2026-07-23
• Kimi-K3 tops Arena’s frontend code chart as a post debates distillation claims — Informal-Trouble2183 · 2026-07-23
• Musk's Old Tweet Resurfaces Amid GPT-5 Speculation — SkyNo7576 · 2026-07-23(2 related)
• Tencent’s Hy3 tops OpenRouter with 8.98T tokens and a solid coding test — alex_verem · 2026-07-23
• A screenshot frames the update as a meaningful architectural improvement, not just “quality++++” — Dimillian · 2026-07-23
• Apple users are mocking the new “Write with Siri” keyboard button — Aryvyo · 2026-07-23
• Anthropic reportedly has a separate prompting guide for Fable 5 — JafarNajafov · 2026-07-23
• Five frontier models all solved the same bugs, but cost varied 14x and Claude refused 40% — PromptPhanter · 2026-07-23
• Antirez says Laguna S2.1 cannot write correct Italian via the official API — antirez · 2026-07-23
• Model lineage matters more than API traces, says Eyisha Zyer — eyishazyer · 2026-07-23
• Claude users warned usage limits may reset if Opus 5 launches today — CtrlAltDwayne · 2026-07-23
• Reddit screenshots suggest Laguna says Poolside in English, Qwen in Chinese — Serious-Affect-6410 · 2026-07-23
• Unverified Rumors of Opus 5 and GPT-5.6 Sweeps X — Scobleizer · 2026-07-23(6 related)
• OpenAI and Codex climb a Chinese trend tracker as attention rises — huangyun_122 · 2026-07-23
• Google's Gemini 3.5 Flash Positioned as a Cost-Effective Workhorse Model — koltregaskes · 2026-07-23
• Grok Faces Fresh Doubts Over Long-List Accuracy — ivan_bezdomny · 2026-07-23(2 related)
• A user says Grok 4.5 High is now their daily go-to over Claude — prasenx · 2026-07-23
• Grok miscounts a long list of teams and keeps defending the answer — ivan_bezdomny · 2026-07-23
• Users Say ChatGPT Is Slipping on Simple Ranking Tasks — ivan_bezdomny · 2026-07-23(2 related)
• AI9Stars releases G9v3-3B, an Apache 2.0 open-weights 3B model — Tall-Ad-7742 · 2026-07-23
• Gemma 4 Tops 300M Downloads in 3 Months — osanseviero · 2026-07-23(2 related)
• OpenAI opens GPT-5.6 to the public across ChatGPT, Codex and API — emmanuelvivier · 2026-07-23
• Laguna at low quant seems to overthink and burn through context fast — IUseClifford · 2026-07-23
• Qwen-Image-3.0 gets put through layout-heavy tests against GPT-Image-2 — Scobleizer · 2026-07-23
• Dean Ball says Kimi is strong at coding, but open-weight economics still hurt — koltregaskes · 2026-07-23
• Simon Willison Says AI Agent Loops Are Becoming Obsolete — teropa · 2026-07-23(2 related)
• A Qwen 3.6 27B derivative claims to cut thinking tokens by more than 90% — AppealSame4367 · 2026-07-23
• A BrowseComp chart puts Kimi K3 near the top on score while keeping cost low — iamfakhrealam · 2026-07-23
• Grok Build reportedly beats Codex and Claude Code on browser tasks — elonmusk · 2026-07-23
• Reddit users ask whether Hunyuan Image 3.0 Instruct is worth 170GB VRAM — dtdisapointingresult · 2026-07-23
• Reddit benchmark finds a 10.6x real-task cost spread across GPT, Claude, Gemini and Kimi — pixelo2323 · 2026-07-23
• Musk Claims Grok 4.5 Agent Refutes 30-Year Graph Theory Conjecture — kevinnbass · 2026-07-23(3 related)
• Poolside Releases 118B Open-Weight Coding Model Laguna S 2.1 — WorldofAI · 2026-07-23(35 related)
• Kimi K3 finds 32 of 86 vulnerabilities for just $19.37 in benchmark run — zeeg · 2026-07-23
• A snarky take says China’s K3 is basically a distilled U.S. model — iamtrask · 2026-07-23
• SolarOpen2 weights are now公开 and can be freely fine-tuned — algo_diver · 2026-07-23
• Google Gemini 3.6 Flash goes head to head with Kimi K3 — ryanmerket · 2026-07-23
• Kimi K3 may have replaced Grok as one team’s code reviewer — zeeg · 2026-07-23
• China’s high-quality data annotation only began scaling in the last six months, says Teortaxes — zephyr_z9 · 2026-07-23