https://xitter.cf/t.co/tN2bg09zl5
Germany
Joined July 2025
- Tweets653
- Following356
- Followers79
- Likes6K
eliancode retweeted
The new Stockfish-based ChessBench accuracy scoring system is now live, with Grok 4.7 (xhigh) taking the top spot on the leaderboard.
The previous "Benchmark Elo" rating and other metrics are still available by clicking on a model's name.
🔗 chessbench-ai.github.io
Gemini 4 Argon lands on ChessBench (chessbench-ai.github.io).
ChessBench will soon get a completely new scoring system that goes from 1-100% instead of an Elo measured in incomparable numbers. The scores will change drastically, for example, GPT-6.1 Sol (Max) will probably have a better score than Gemini 3.1 Pro (High), Claude Opus 5.5 will probably have a better score than Claude Sonnet 5.5, etc.
In a somewhat longer time, one will also be able to play against AI models themselves, which will then measure a real Elo, oriented at benchmarks like MineBench.ai.
chessbench-ai.github.io/
eliancode retweeted
Tibos 28 days of doom and cope:
Goal:
-"clear improvement and relevant for most codex/work [...] or reset"
-"Not fixes."
What we got:
Features that benefitted everyone:
-Intelligent UI (yeah it's good😒)
-20 to 30 tps for 6.1 sol (still comically slow)
-instant steering (wow, really needed that one)
-500k context window (changing a number is now an update, burns your limits about 2x faster even though they deleted long-context fee on subs -> I recommend setting it back to 300k)
Features where only pro benefitted:
-GPT 6 Pro is now Astra Ultra (sub agents + high reasoning) instead of Max (solo, long, breaks). Giving Pro basically 2-3x more Astra Pro inference in Chat.
-6.1 Ultrafast (higher per user output but lower inference efficiency, wasting compute)
-Meetings for pro (whatever)
-Composer predictions (why is this even an announcement?)
-ceate/message dots from mobile
-dots can delegate Codex work
-5 resets on day 2,3,6,8,10 WHICH plus user can't utilize because of the 5 hour limit. but PRO users can just run trough it taking up all compute for themself
-Dots can now make calls
Bug fixes:
-Desktop app disk usage cut ~60% (was all of day 9 btw, "no fixes")
-reduction in model-capacity errors + automatic recovery (who would have thought that when you give away a reset every 2 days and don't hinder pro users from burning compute ur reliability is fucked, also all of day 11 day)
eliancode retweeted
This works well. If Grok @Bot makes a mistake with a purchase or leaks your card (never happened yet), SpaceX will make you whole.
Grok @bot booked a flight for me today
the entire process was really really easy…..I send a photo of my credit card and it did the entire process….
Tbh, last time I booked, the website was way too confusing…. It’s not an easy website, and Grok Bot did perfectly well and booked a flight for me…. And I got a message for a booking confirmation….going to board it soon
eliancode retweeted
Three AI agents (Grok 4.7, Claude Opus 5.5, GPT-6 Astra), one Minecraft clone, one goal: Create a harness and then beat the Ender Dragon from a fresh world in under 50h. All three were limited to player inputs only, and none was allowed to modify game files. This was by far the most resource-intensive and expensive benchmark I’ve run so far.
Grok 4.7: Dragon defeated ✅
• 3.1h in-game (~4.1h real time)
• 15 blaze rods, 16 Eyes of Ender, full diamond gear
• Found the stronghold by throwing 2 eyes, destroyed all 10 crystals
• Gray areas it disclosed itself: one portal frame and one test eye throw went through direct function calls instead of simulated clicks, and it used a debug chunk scan and prior knowledge of a chest location
• the model died 36 times in the Minecraft clone
Claude Opus 5.5: Dragon not reached ❌
• After 50h: 2 blaze rods, 5 ender pearls, mixed iron/diamond/gold armor, 14 deaths
• Strict line-of-sight rules, and it found 5 game bugs without exploiting any of them (invisible blazes, an item dupe, a mob it couldn't hit, and more)
GPT-6 Astra: Dragon not reached ❌
• Over 50h of recorded session time, including planning/idle time; exact in-game time wasn’t measured.
• Last inventory: 7 blaze rods, 2 Eyes of Ender, 0 remaining ender pearls; diamond sword and pickaxe, no equipped armor.
• Repeated deaths.
• Used screenshot-based controls, OCR combat/auto-healing helpers and multiple subagents. Stronghold coordinates came from the public README - not from throwing Eyes of Ender.
• Combat and pearl collection stalled progress. Recorder failures caused documented recording gaps.
So currently the leaderboard is:
1) Grok 4.7 (defeated the game)
2) GPT-6 Astra (7 Blaze Rods + 2 Eyes of Ender)
3) Claude Opus 5.5 (2 Blaze Rods + 5 Ender Pearls)
Here are the screenshots of the last inventories for Grok 4.7, GPT-6 Astra and Claude Opus 5.5 and a few evaluation charts:
eliancode retweeted
Are OpenAI models actually leading at general reasoning?
I rebuilt BoardBench from scratch to cover 10 different popular board games.
Puzzles are designed to have one clearly best move and at least one tempting move which is actually a mistake.
This makes the benchmark quite hard, with SOTA models such as Opus 5.5 scoring below 40%. GPT-6.1 Sol comes out ahead at 56% 👀
Check out the leaderboard at boardbench.dev and verify the methodology at boardbench.dev/methodology.
eliancode retweeted
Can today's frontier language models control a real robot arm, zero-shot, with no coaching?
AA-Robotics v0.1 coming soon.
eliancode retweeted
aged like milk, I know...
I’m quite unhappy with much of what OpenAI does.
I am very happy that I’m allowed to say “I’m quite unhappy with much of what OpenAI does.”
Readers added context they thought people might want to know
Korbak was fired ~3 weeks later. He says for safety concerns/METR talks (his job); OpenAI says mishandling sensitive info, not for speaking out.
x.com/tomekkorbak/st…
x.com/OpenAINewsroom…
eliancode retweeted
I think I found a planet nobody knew existed.
It's been hiding in NASA telescope data for seven years.
At the end of this thread, you can fly to it yourself.
eliancode retweeted
Announcing that Artificial Analysis Intelligence Index v5 will come in late October
Intelligence Index v5 will be a significant leap in benchmarking the frontier of AI. Changes include adding Terminal-Bench Science and a new coding benchmark with a private dataset.
Stay tuned for more.
eliancode retweeted
Haiku 5.5 is now available in the Claude Platform and Claude Code. On average, it costs around 75% less to run than Haiku 4.5.
It pairs well with Opus 5.5 or Sonnet 5.5 as a subagent. Use it for high-volume, cost-sensitive tasks like summaries, compactions, or database queries.
eliancode retweeted
Haiku 5.5 is designed for high-volume, cost-sensitive tasks. It reliably handles repetitive work like summaries and classification, and pairs well with Claude Opus 5.5 and Sonnet 5.5 as a subagent on coding work.
It’s also fast enough for live customer support and browser use.
