PERPLEXITY : Grok 4.5 scored the highest on WANDR with Perplexity Computer, landing at the same price range as recently announced custom GLM-5.2 with Advisor.
ECONOMY
-
AI Accelerates Existing Economic and Strategic Pressures on the Press
By
–
AI is probably not the cause of every problem. Above all, it accelerates the consequences of weaknesses the press has faced for years: strained business models, sometimes questionable strategic decisions, and insufficient investment. The disappearance of a publication [Traduit de FR vers EN]
-
GPT-5.6 Sol User Reports Rapid Five-Hour and Weekly Quota Exhaustion
By
–
Another reset, which is unfortunately necessary as well. Unfortunately, it is becoming increasingly clear that the token burn rate is simply insane. I have never come close to hitting the 5-hour and weekly limits this quickly on the Pro tier as I have with 5.6 Sol. I mostly use
-
Hermes Agent Uses GPT-5.6 Sol for Robinhood Trading Research
By
–
-

Grok 4.5 and Muse Spark Lead Legal Agent Benchmarks at Lower Cost
By
–
Legal Agent Benchmark (
@harvey
): Grok 4.5 and Muse Spark 1.1 achieve SOTA at significantly lower cost and latency: -
Orchestrators Account for Much of Long-Horizon Agent Task Costs
By
–
60% of the cost of a task is the orchestrator. So it’s important to have a cost-efficient intelligent model as your orchestrator even if your objective is to minimize cost of your long horizon agent trajectory.
-
Sol Ultra User Reports Spending Two Billion Tokens Overnight
By
–
Yesterday, GPT-5.6 Sol Ultra became generally available. Today, I'm sharing that I accidentally spent 2b tokens overnight and I didn't even solve a 50-year-old math conjecture. This is extremely cool though, skill issue on my part clearly
-
High Compute Spending Does Not Guarantee Strong AI Models
By
–
Not everyone will succeed in the model business.
— Cohere (@cohere) 10 juillet 2026
"You've seen a lot of companies spend a huge amount on compute and then not end up with a good model." – @nickfrosst pic.twitter.com/vSYwEYnpd4Not everyone will succeed in the model business. "You've seen a lot of companies spend a huge amount on compute and then not end up with a good model." – @nickfrosst
-

LangGraph and Perplexity Produce Cited VC Memos From Parallel Research
By
–
This agent drafts a cited VC investment memo in ~90 seconds for $0.40. 4 LangGraph nodes powered by the @Perplexity_AI Agent API research financials, product, and market in parallel A tool-less synthesizer writes the memo from only what they found Deep dive:
-
Muse Spark 1.1 Targets Opus-Level Tasks at Lower Cost
By
–
muse spark 1.1 can be even better than opus 4.8 at 20% of the cost https://t.co/gMfZ7884iE
— Alexandr Wang (@alexandr_wang) 10 juillet 2026muse spark 1.1 can be even better than opus 4.8 at 20% of the cost
