⚔ GLM-5.3-Flash Aims Straight at the Post-Hike DeepSeek Zhipu's GLM-5.3-Flash — revealed this week as the anonymous "Ox Alpha" — has been open-weighted and priced at roughly one-tenth of GLM-5.3. Much of the early discussion compares it to DeepSeek's V4 Flash, which recently raised prices. Zhihu contributor 起步十档, who ran Ox Alpha inside real workflows before the reveal, gives a practitioner's verdict in one line: it is built to kill the post-hike DeepSeek. His case rests on three legs — performance, token efficiency, and an architecture change that makes the price possible. 1️⃣ It clears the bar for long-horizon work Official scores put Flash between Grok 4.6 and GLM-5.3, and clearly ahead of DeepSeek V4 Flash. In the author's own testing, its frontend ability roughly matches an early gray-test build of DeepSeek V4 Pro, while its backend is noticeably weaker than GLM-5.3 — but still usable on long-horizon tasks as long as the connection holds. His rule of thumb: any model past the Claude Opus 4.6 line is workflow-ready for long tasks. Beyond that, differences come down to reasoning style and accuracy, not viability. One caveat he flags: during the anonymous test the deployment was unstable, and some believe it served a mid-training checkpoint rather than the final model. 2️⃣ The real weapon: token efficiency Comparing peak API prices against the post-hike DeepSeek V4 Flash, the author notes cached input is actually 2x more expensive, while regular input and output sit at roughly 26% of DeepSeek's price. Since cached input is often the bulk of the bill, he wants real-world tests before calling a winner on price alone. But his own usage points the same direction. In one to two hours of real work — reading and writing files, running tests — Ox Alpha burned barely over 100K tokens. He estimates DeepSeek would need 250-300K for the same workload. His prediction: same tasks, run on both APIs, will come out cheaper on Flash — with clearly better performance. 3️⃣ His unexpected advice: skip the Coding Plan The plan only triples your quota for Flash. Using Zhipu's own best-case math — maximum usage, off-peak hours, the official 0.8x API-equivalent rate — the plan works out to about 53% of pay-as-you-go API cost. Since that scenario is already extreme, he concludes the API is the better deal for almost everyone. 4️⃣ The architecture change behind the price From GLM-5 through 5.3, Zhipu used DSA — essentially an optimized full attention — which costs more than DeepSeek's CSA/HCA, Kimi's KDA, or Qwen's linear-global hybrid. That is why GLM used to be pricier than larger DeepSeek models. Flash is the first GLM to switch to linear attention plus an HCA-like compressed attention, trained with the HCA recipe as well. That brings it in line with mainstream domestic practice — and the price fell accordingly. The author expects a future GLM-5.5 can scale up without costing much more than 5.3. On top of that, Zhipu added native multimodality and leaned on domestic compute, which he reads as the reason for the generous free quotas during the anonymous test. 5️⃣ Zhipu is still the team to beat The author's closing line is unambiguous: Zhipu remains, in his words, the number-one Chinese model company. That is his judgment, not a benchmark result — but the cost argument underneath it is now easy to check yourself. 🔗 Key links: Official announcement: Open weights (MIT): 🔗 Full Reading: #GLM# #Zhipu# #DeepSeek# #LLM# #AIInference# #TokenEfficiency# #OpenWeights#

Micron just laid out its AI memory roadmap through 2030. Token efficiency = Bandwidth + Energy Efficiency + Capacity • HBM (JEDEC + Custom) → HBM4 to HBM6 • vHBM (Vertical 3D HBM) starts 2028 • vEIM (Vertical Edge Inference Memory) targets 2TB/s & <1pJ/bit Vertical stacking is coming. Memory wall is real. #HBM# #Micron# #AI# #Semiconductors# #vHBM#

Sam Altman reveals the benchmark that may matter more than model IQ: 54% better token efficiency on agentic coding "5.6 Sol, I think, is not only the best model in the world for most people." "It's also much more efficient than other models out in the world." "So it's 54% more token efficient on agentic coding tasks and also as good or better as the other best models out there." "And we're really seeing people now start to care about efficiency, understand their spend, get a great ROI." "So this is a great step forward for us." The model race is moving from leaderboard intelligence to cost per completed task. The same agentic work at radically lower token cost changes product margins, rate limits, and how much autonomy an enterprise can afford. The next moat may not be the smartest model. It may be the system that turns each dollar of inference into the most reliable work. - Sam Altman (@sama), CEO of OpenAI, on @CNBC

just finished an hour of sexual harassment course and was thinking we should have a token harassment course given a task, what's the right model/prompt/context teach employees token efficiency, context window management, model selection and prompting skills

Anthropic has proven 3 things: - Taste strongly influences whether people will like a model - People don't care about token efficiency if the model gives good results - Understanding intent well creates a better user experience and less prompt iterations

GPT-6 Astra is a very good model in a lot of ways but i think the token efficiency might be the most impressive part compared to GPT-5.6 sol, Astra gets significantly better performance while using roughly the same, or sometimes even fewer, tokens at higher reasoning levels