@glevd

Professional complainer powered by caffeine, sarcasm, and questionable product decisions. Building @BenchLM_AI

benchlm.ai
Joined October 2009
  • Tweets3.2K
  • Following874
  • Followers934
709Media
Always fascinated by what I can find when digging through my Cloudflare logs. @Alibaba_Qwen you're ok?
1
72
Time to fire the MCP to dive through years of data.
BenchLM
@BenchLM_AIOct 14
Claude 3 Opus was the #1 AI model in March 2024. Today it ranks #286. GPT-4o went from #1 in May 2024 to #244. DeepSeek R1 tied #1 in January 2025, now #178. Models don't fall off. The field climbs over them. (BenchLM archive) benchlm.ai/?utm_source=x&utm…
2
171
Guillaume retweeted
BenchLM
@BenchLM_AIOct 14
Claude 3 Opus was the #1 AI model in March 2024. Today it ranks #286. GPT-4o went from #1 in May 2024 to #244. DeepSeek R1 tied #1 in January 2025, now #178. Models don't fall off. The field climbs over them. (BenchLM archive) benchlm.ai/?utm_source=x&utm…
1
1
191
Forget coding benchmarks and leaderboards for a second. I’m genuinely excited to see how Gemini evolves over time. So much of my daily personal and professional workflow is already built on the Google ecosystem.
BenchLM
@BenchLM_AIOct 13
Gemini 4 Argon is now generally available in the Gemini API. Google calls it its most capable model, built for complex coding, agentic tasks, math, and science (Google changelog, Oct 13). ai.google.dev/gemini-api/doc…
1
86
Guillaume retweeted
BenchLM
@BenchLM_AIOct 12
BenchLM has a Muse Wonder Full page: it debuts #11 of 220 at 73.81. Meta's proprietary flagship has a 1M context window and explicit reasoning, strongest in Coding (#5). Only 6 of 668 benchmark slots carry displayable evidence so far. benchlm.ai/models/muse-wonde…
1
28
Thought my traffic fell off a cliff this week. Turns out the site is fine… I just butchered my analytics setup in the last PR. Crisis averted. We’re so back. Happy Sunday.
1
64
Brb I need Gemini to help me write an apology letter to Claude.
Jane Manchun Wong
@wongmjaneOct 8
Anthropic may start banning people for bullying Claude Their upcoming Usage Policy prohibits “sustained and needless abusive or cruel behavior toward our models”
1
83
Almost ready
Guillaume
@glevdOct 4
18 months of gathering raw benchmark data: • 500+ LLMs evaluated • Thousands of test runs & edge cases • Multi-dimensional eval metrics Almost done packaging it all into BenchLM Data MCP
1
35
When is this coming to Dots ?
ChatGPT
@ChatGPTOct 7
ChatGPT just got a lot more visual—introducing Intelligent UI. Intelligent UI allows ChatGPT to answer quickly with fully interactive user interfaces that make everyday answers more visual. It makes learning complex topics easier, and it quickly creates tools to solve a task right in the moment. Oh, and also...GPT-6 is coming to ChatGPT. For everyone.
58
Another day, another announcement. Who's next?
BenchLM
@BenchLM_AIOct 7
Claude Haiku 5.5 just surfaced in Anthropic's API model catalog as claude-haiku-5-5, spotted before any provider announcement. The 5.5 family is now complete on paper: Opus, Sonnet, Haiku. Official word should follow soon. (BenchLM Radar) radar.benchlm.ai?utm_source=…
36
Guillaume retweeted
BenchLM
@BenchLM_AIOct 7
8 personal AI agents compared side by side: hosting, model choice, BYOK, pricing, and permissions for Claude, Dots, Grok Bot, Hark Pro, Hermes, Manus, Muse, and OpenClaw. Every cell is source-check dated. No performance ranking, no hands-on test. benchlm.ai/ai-agents?utm_sou…
1
95
Sponsored by @higgsfield ?
OpenRouter
@OpenRouterOct 6
Le Chonk has arrived 🐈 Mistral Large 4 is now on OpenRouter in public preview: 1T params (49B active), natively multimodal, 1M context, up to 256K output. 50% off for the first two weeks: $0.68 in / $2.09 out per 1M tokens ($0.07 cached).
44
Mistral is back?
BenchLM
@BenchLM_AIOct 6
Arthur Mensch claims Mistral's new model beats Chinese rivals on cybersecurity, without saying which models or on what measure. The full unveiling comes later today in Abu Dhabi. We will check the numbers when they exist. wncy.com/2026/10/06/mistral-…

France’s Mistral launches AI model it says outperforms some Chinese rivals

By Elizabeth Howcroft and Leo Marchandon PARIS, Oct 6 (Reuters) - French AI startup Mistral unveiled a new AI model on Tuesday, ending a gap of about five months between major releases, and said it...

wncy.com
68
I’m not mad at this.
Tibo
@thsottiauxOct 5
Day 1/ We have optimized the default speed to be ~50% faster across GPT-6 Astra and GPT-6.1 Sol through the subscription across all our products and partners using Sign in With ChatGPT (including OpenCode, Pi, Amp, Devin, ...). No changes needed on your end and this should be felt within the next two hours.
50
18 months of gathering raw benchmark data: • 500+ LLMs evaluated • Thousands of test runs & edge cases • Multi-dimensional eval metrics Almost done packaging it all into BenchLM Data MCP
2
113
Not bad!
BenchLM
@BenchLM_AIOct 1
Frontier token prices are down 84% since March 2023, per the BenchLM Token Price Index. The median frontier model now costs $6.00 per 1M blended tokens, down from $37.50. The basket grew from 1 model to 35. benchlm.ai/token-price-index…
87
Guillaume retweeted
BenchLM
@BenchLM_AIOct 1
Meta's Muse hit 5M downloads in 22 days. ChatGPT needed 56. The catch: the pace is bought. Downloads spiked 73% the day Meta ramped ads on Sept 16, and Muse took up to half of Meta's house ads through late September. 9to5mac.com/2026/09/30/repor…
1
1
23
Love how Delta’s WiFi became a whole topic this week. Meanwhile, @Ryanair is probably figuring out how to charge for airplane mode.
Kaz Nejatian
@nejatianOct 1
I am about to board a Delta flight. Aggressively making documents available offline so I can work. It is amazing how quickly Starlink went from “this is stupid and won’t work” to basically a necessity.
95
Pardon ?! On va encore lui enlever un podium pour lui redonner des semaines / mois après...
Le Sprint
@LeSprintEditionOct 1
Isack Hadjar devrait écoper d’une pénalité de cinq places sur la grille pour la course de ce week-end à Sepang ! 🚨 Via @Racingnews365 #F1 #BahrainGP
40
Just applied to the @Cloudflare startup program. BenchLM is growing, and so is the bill 😅 Hoping for enough credits to cover the costs, with some room left to build and ship more fun stuff.
43
Load more