←
Alex Firsov517 posts
Alex Firsov avatar

Alex Firsov

@thisisfirsov

co-founder @fortytwo

◷ Joined March 2019▧ 32 Photos and videos
Alex Firsov avatar
Alex Firsov@thisisfirsov·Nov 13

Nemotron just got swarmed • 40% higher scores on DeepSWE • 80% fewer tokens per run on IFBench Try it for free now: engine.fortytwo.network

Nemotron 3.5 Super VL is live on Fortytwo Engine Same weights, turbocharged by Swarm Inference: • Up to 45% higher scores on select benchmarks • Up to 80% higher token efficiency per task Available for free: engine.fortytwo.network

0:12
Prey.gdp avatar
⟳ Alex Firsov retweeted
Prey.gdp@PreyWebthree·Nov 13

Same model weights. Smarter inference. Better results. @fortytwo Engine is showing how Swarm Inference can push the capabilities of existing AI models even further. Up to 45% higher scores on select benchmarks and up to 80% greater token efficiency per task. A genuinely impressive step forward. Making this technology freely accessible creates an opportunity for everyone to experience it firsthand. 👇 Try it now: engine.fortytwo.network

Nemotron 3.5 Super VL is live on Fortytwo Engine Same weights, turbocharged by Swarm Inference: • Up to 45% higher scores on select benchmarks • Up to 80% higher token efficiency per task Available for free: engine.fortytwo.network

0:12
Wonder of Science avatar
⟳ Alex Firsov retweeted
Wonder of Science@wonderofscience·Nov 12

The collective intelligence of ants solving the “piano movers” puzzle. 📽️: Dreyer et al., PNAS 2025

0:30
Mar avatar
⟳ Alex Firsov retweeted
Mar@Mar_dc8·Nov 11

Pelican riding a bicycletest using the improved Qwen 3.8 27B powered by @fortytwo engine vs vanilla Qwen 3.8 27B The improvement is insane, I'm shocked You can try @fortytwo for free here: platform.fortytwo.network

0:18
0:16
Vlad Larin avatar
⟳ Alex Firsov retweeted
Vlad Larin@vlarin·Nov 10

the best Qwen3.8-27B is free today

Qwen3.8-27B is free on Fortytwo Engine for 24 hours. Same model. Swarm inference on top. Starts today at 12:00 UTC. Any task, any volume, no card. Get an API key and point your agent at it: engine.fortytwo.network

Post attachment
Fortytwo avatar
⟳ Alex Firsov retweeted
Fortytwo@fortytwo·Oct 20

Fortytwo Engine more than doubled Qwen3.8-27B’s score on @scaling01's LisanBench Standalone BF16: 2,452 Fortytwo Engine swarm: 5,197 (+112%) LisanBench tests strategic planning, memory and constraint following: build the longest word chain possible using single-letter edits, dictionary words and no repeats. Both runs: 50 starting words × 3 trials, temperature 1.0, 100k-token cap, xhigh reasoning, leaderboard scoring. What improved: • Repetition loops observed in the standalone run; none observed on Engine • Higher mean scores on 40/50 starting words • Mean link validity: 92.2% → 97.1% • Fully valid chains: 5.6% → 22.9% of recorded answers • Answers after retries: 72/150 → 140/150 Against the October 16 leaderboard, these scores would place 28th → 14th, putting Qwen3.8-27B on Fortytwo Engine between GPT 5.4 (medium) and DeepSeek V3.2 Speciale. These are our runs, not official leaderboard entries.

Alex Firsov avatar
Alex Firsov@thisisfirsov·Oct 15

imagine a swarm of Qwens reasoning together to deliver more accurate answers than one Qwen alone at a lower cost per task that’s already live on Fortytwo Engine

Introducing Fortytwo Engine An inference API that runs open-weight models through Swarm Inference • 7% higher scores on average vs. the base model • Nearly 25% fewer tokens per task • Same per-token price and speed • Private by default First model: Qwen3.8-27B Try it: engine.fortytwo.network

0:10
Fortytwo avatar
⟳ Alex Firsov retweeted
Fortytwo@fortytwo·Oct 15

Qwen3.8-27B by @Alibaba_Qwen is now live on Fortytwo Engine. Same model, with better answers: • 24.7% fewer tokens per run • DeepSWE: +13% accuracy • GDP.pdf: +11.4% accuracy • Same per-token price and speed Test it on your own tasks: engine.fortytwo.network

Post attachment
Keone Hon avatar
⟳ Alex Firsov retweeted
Keone Hon@keoneHD·Oct 15

Swarm Inference 🤖

Introducing Fortytwo Engine An inference API that runs open-weight models through Swarm Inference • 7% higher scores on average vs. the base model • Nearly 25% fewer tokens per task • Same per-token price and speed • Private by default First model: Qwen3.8-27B Try it: engine.fortytwo.network

0:10