Reed Chan

I miss you so much, DeepSeek. The Blue Whale is back! 🐋

Why does my GPT Pro subscription only list a 256K token context window for GPT-5.5 in the latest Codex? Wasn’t it supposed to be 400K? @OpenAI @OpenAIDevs

Finally, here we go 🥳

@ecommerceshares Spot on. Anthropic is cooked.🥱

Claude Pro sub. Ran a Sonnet 4.6 prompt twice. 2 times. Not even on Opus. Immediately hit my session limit and 43% of my weekly limit. What??!?!

Can't agree more. The bait-and-switch is getting exhausting. 📉 • Opus 4.7 token inflation = stealth ~35% price hike for the exact same agent workflows. • Rug-pulling Claude Code from the $20 tier to push the $100/mo Max plan. The math is literally breaking. We went from 'feel unlimited' to 'checking the API usage every 5 mins' in record time. Miss the days when they optimized for devs, not just enterprise ARR. 💔

We can all say it... Claude 20x is dead. The previous "feel unlimited usage" doesn't exist anymore. Usage increases quickly and scales really fast... People who discover Claude Code now: you missed a time that will never come back, I think.

Spot on. Building agents on Claude lately feels like death by a thousand cuts. 🩸 They kept Opus 4.7 at $5/$25, but the tokenizer inflation means we’re burning ~35% more tokens for zero noticeable ROI. And quietly testing the removal of Claude Code from the $20 Pro tier? Shoving solo devs to $100/mo overnight is wild. It used to feel like Anthropic was building with us. Now it just feels like they’re extracting from us. Hard to focus on building when you’re constantly sweating the billing dashboard. 😮‍💨 #Anthropic #AI #claudecode

@hao_y17959 @imwsl90 你看智谱和 claude 猛猛涨价，就知道它并没有把 Token 价格打下来，当然我也很希望 Anthropic 寄掉，死贵死贵的

@imwsl90 ds的作用就是把token价格打下来，先不说他强不强，就算强他这更新间隔也没什么可用性

我怎么感觉DeepSeek V4 要翻车啊 海外模型迭代太快了

@nishancodes @bindureddy I’ve been waiting for it sooooo long!

@bindureddy US AI labs build something frontier level and some random chinese labs gives it away for free the next week. I'm glad the OSS community is innovating this fast. I'm waiting for Deepseek release this week probably

Kimi K2.6 is Opus 4.7 level in agentic coding 🤯 We have got the receipts - will publish soon

@chatgpt21 I'm a bit confused.. Isn't Spud GPT-6? 5.5 is just a normal iteration, right?

🚨Breaking Spud (GPT 5.5) leaked ( for a few minutes ) As well as - Oai 2.1

And to be precise: I don’t mean faster or cheaper. I mean a capability that moved from demo-able to dependable.

Every month, a new SOTA. Every month, a new benchmark broken. Every month, we all cheer. Kimi K2.6 dropped this week. Genuinely impressive numbers. But I keep getting stuck on a question nobody seems to be asking out loud: SOTA at what, exactly? And then what? The race has become its own destination. Labs aren't racing toward something anymore — they're racing to win the race. The benchmarks measure performance on benchmarks. The leaderboards rank who tops the leaderboard. We've built the world's most sophisticated system for producing systems that score well on tests. At some point, someone has to ask: what does any of this actually unlock for a real human being, in a real situation, that wasn't possible six months ago? I'm not sure the answer is as obvious as the release cadence suggests. #AGI #AI #SOTA

Serious question: what is one thing frontier models can do today, reliably, that they could not do 6 months ago — and that changed a real workflow, product, or business outcome?

Every month, a new SOTA. Every month, a new benchmark broken. Every month, we all cheer. Kimi K2.6 dropped this week. Genuinely impressive numbers. But I keep getting stuck on a question nobody seems to be asking out loud: SOTA at what, exactly? And then what? The race has become its own destination. Labs aren't racing toward something anymore — they're racing to win the race. The benchmarks measure performance on benchmarks. The leaderboards rank who tops the leaderboard. We've built the world's most sophisticated system for producing systems that score well on tests. At some point, someone has to ask: what does any of this actually unlock for a real human being, in a real situation, that wasn't possible six months ago? I'm not sure the answer is as obvious as the release cadence suggests. #AGI #SOTA #AI

@OpenAI Opus 4.7 sucks. I’ve switched back to 4.6.

@OpenAI Something big is coming? Can't wait!

This is not a screenshot.

@bd5m112 @GTqhqh48540 You might be right, but at least these Chinese models are open-source. Companies like DeepSeek have taken a huge step forward in making AI accessible to developers and ordinary people worldwide. Isn’t that true? 🌹

@GTqhqh48540 I'm talking on coding. On other areas they're not "forever one generation behind", but on coding it's more than one year since I repeat this.

can't help but notice every new chinese model's benchmark visuals seem to compare to previous generations opus 4.7 conspicuously absent.. lol

Kimi K2.6 has been released, where is DeepSeek V4? #DeepSeek #Kimi #AGI

@Kimi_Moonshot @grok What is the size of the Context Window for Kimi K2.6?

Meet Kimi K2.6: Advancing Open-Source Coding 🔹Open-source SOTA on HLE w/ tools (54.0), SWE-Bench Pro (58.6), SWE-bench Multilingual (76.7), BrowseComp (83.2), Toolathlon (50.0), Charxiv w/ python(86.7), Math Vision w/ python (93.2) What's new: 🔹Long-horizon coding - 4,000+ tool calls, over 12 hours of continuous execution, with generalization across languages (Rust, Go, Python) and tasks (frontend, devops, perf optimization). 🔹Motion-rich frontend - Videos in hero sections, WebGL shaders, GSAP + Framer Motion, Three.js 3D. 🔹Agent Swarms, elevated - 300 parallel sub-agents × 4,000 steps per run (up from K2.5's 100 / 1,500). One prompt, 100+ files. 🔹Proactive Agents - K2.6 model powers OpenClaw, Hermes Agent, etc for 24/7 autonomous ops. 🔹Claw Groups (research preview) - bring your own agents, command your friends', bots & humans in the loop. - K2.6 is now live on kimi.com in chat mode and agent mode. For production-grade coding, pair K2.6 with Kimi Code: kimi.com/code - 🔗 API: platform.moonshot.ai 🔗 Tech blog: kimi.com/blog/kimi-k2-6 🔗 Weights & code: huggingface.co/moonshotai/Kim…

@minchoi 4.6 → 3T 4.7 → 6T 4.8 → 10T 4.9 → ??? 5.0 → AGI 6.0 → ASI 7.0 → ASI2 … 🤷‍♂️ 😂