Day 7 · 2026-11-17
35 launches checked147 claims quoted word for word179 user reports linked10 verified fixes linked40 corrections publishedLive 25 days5 of 5 founding sponsor spots left
AI launches, checked against real usersLast re-stamped 17 Nov 2026

AI models / Compare

Gemini 4 Hydrogen vs LFM3: which delivered after 7 days?

LFM3 is stamped Delivered; Gemini 4 Hydrogen is still on the record, unstamped, so this comparison isn't settled yet.

Gemini 4 HydrogenLFM3
VerdictOn recordDelivered
Claims held12
Claims mixed01
Claims failed00
Claims unclear30
User reports38
Launched12 Nov 202627 Oct 2026
Last stamped13 Nov 202613 Nov 2026

No. 032 · Google DeepMind

Gemini 4 Hydrogen

Day 5 of 7
On recordDay 7 verdict

Day 1: it is live across Google's products, but the benchmark and speed claims are Google's own and independent hands-on reports are only starting.

Launched
12 Nov 2026
Stamped
13 Nov 2026
Claims
✓1 held · ~0 mixed · ✗0 failed · ?3 unclear
Reports
3 (2 praise · 0 problem · 1 mixed)

What they claimed

“Hydrogen surpasses Argon on the Artificial Analysis Intelligence Index at its default high thinking level (51 vs. 50”
? Unclear
“finishing tasks 2.5 times faster at its high default setting”
? Unclear
“Starting today, Hydrogen is rolling out globally across our developer, enterprise, and consumer surfaces”
✓ Holds
“free users will get access to Gemini 4 Hydrogen via auto model selection when it’s the appropriate model for the prompt”
? Unclear

Full receipt

No. 013 · Liquid AI

LFM3

DeliveredDay 7 verdict

Tiny, local and long-context exactly as promised; the 'beats our bigger model' line is still Liquid's own, and it fails at coding.

Launched
27 Oct 2026
Stamped
13 Nov 2026
Claims
✓2 held · ~1 mixed · ✗0 failed · ?0 unclear
Reports
8 (1 praise · 1 problem · 6 mixed)

What they claimed

“a new tri-hybrid architecture and training recipe allow LFM3-1.7B to outperform our previous LFM2.5-2.6B while using 35% fewer parameters.”
~ Mixed
“the text model fits in about 1 GB at 4-bit precision.”
✓ Holds
“256K context window for the text model.”
✓ Holds

Full receipt

Both stamps come from the public user reports linked on each receipt, under the same method. Verdicts are opinion and are never paid for.

Keep comparingall AI models