Google releases Gemma 4 open models • Hacker News Board

Re: Google releases Gemma 4 open models

#371

Post by hacker_homie » Fri, Apr 03, 2026, 2:33 AM UTC

logicallee wrote: post #117

canyon289 wrote: post #51

Hi all! I work on the Gemma team, one of many as this one was a bigger effort given it was a mainline release. Happy to answer whatever questions I can

Do any of you use this as a replacement for Claude Code? For example, you might use it with openclaw. I have a 24 GB integrated RAM Mac Mini M4 I currently run Claude Code on, do you think I can replace it with OpenClaw and one of these models?

Honestly for that [Qwen3-Coder-Next-GGUF](https://huggingface.co/unsloth/Qwen3-Coder-Next-GGUF)

still seems to be the best in class.

I am testing the Gemma4 now I will update this comment with what I find.

Re: Google releases Gemma 4 open models

#372

Post by hacker_homie » Fri, Apr 03, 2026, 2:35 AM UTC

canyon289 wrote: post #51

Hi all! I work on the Gemma team, one of many as this one was a bigger effort given it was a mainline release. Happy to answer whatever questions I can

Could you please work on tool calling gemma still seems very bad at it.

Re: Google releases Gemma 4 open models

#373

Post by dakolli » Fri, Apr 03, 2026, 2:40 AM UTC

originalvichy wrote: post #14

The wait is finally over. One or two iterations, and I’ll be happy to say that language models are more than fulfilling my most common needs when self-hosting. Thanks to the Gemma team!

[flagged]

Re: Google releases Gemma 4 open models

#374

Post by chrischavez » Fri, Apr 03, 2026, 2:41 AM UTC

Went through the official blog and the developers post, no mention of TurboQuant anywhere. Google's own research team tested it on Gemma models for KV-cache compression to 3 bits, so it's surprising it's not mentioned in this release. Anyone know if it's baked in already or if we'd need to apply it ourselves? Would love to run the 26B MoE locally as a daily driver.

Re: Google releases Gemma 4 open models

#375

Post by nl » Fri, Apr 03, 2026, 2:45 AM UTC

Gemma-4-E4B-it scored 15/25 on my https://sql-benchmark.nicklothian.com/#all-data (agentic SQL generation).

The naming is a bit odd - E4B is "4.5B effective, 8B with embeddings", so despite the name it is probably best compared with the 8B/9B class models and is competitive with them.

Qwen3.5-9B also scores 15/25 in thinking mode for example. The best 9B model I've found is Qwen3.5-9B-Claude-4.6-Opus-Reasoning-Distilled-v2 which gets to 17/25

gemma-4-E2B (4bit quant) scored 12/25, but is really a 5B model. That's the same as NVIDIA-Nemotron-3-Nano-4B which is the best 4B model I've found (yes, better than Qwen 4B).

That's a great score for a small model.

Re: Google releases Gemma 4 open models

#376

Post by TGower » Fri, Apr 03, 2026, 2:55 AM UTC

canyon289 wrote: post #51

Hi all! I work on the Gemma team, one of many as this one was a bigger effort given it was a mainline release. Happy to answer whatever questions I can

Any chance of Qualcomm NPU compatible .litertlm files getting released?

Re: Google releases Gemma 4 open models

#377

Post by khimaros » Fri, Apr 03, 2026, 3:19 AM UTC

c0wb0yc0d3r wrote: post #354

yags wrote: post #348

Earlier quoted context omitted.

We (LM Studio) found the bug with the 31B model and a fix will be going out hopefully tonight

I am not deep in this world. What does it mean when you (LM Studio) fixed a bug in a model Google released?

llama.cpp also fixed some chat template issues this afternoon. could be related.

Re: Google releases Gemma 4 open models

#378

Post by nostrebored » Fri, Apr 03, 2026, 3:43 AM UTC

WarmWash wrote: post #58

ceroxylon wrote: post #16

Even with search grounding, it scored a 2.5/5 on a basic botanical benchmark. It would take much longer for the average human to do a similar write-up, but they would likely do better than 50% hallucination if they had access to a search engine.

Even multimodal models are still really bad when it comes to vision. The strength is still definitely language.

Training for tasks still works petty well, but “vision” is a super broad domain and most seem optimized for OCR and screen processing (which have verifiable outputs and relatively straightforward data generation)

Re: Google releases Gemma 4 open models

#379

Post by FullyFunctional » Fri, Apr 03, 2026, 3:44 AM UTC

nateb2022 wrote: post #80

kpw94 wrote: post #45

Earlier quoted context omitted.

Wild differences in ELO compared to tfa's graph: https://storage.googleapis.com/gdm-deepmind-com-prod-public/... (Comparing Q3.5-27B to G4 26B A4B and G4 31B specifically) I'd assume Q3.5-35B-A3B would performe worse than the Q3.5 deep 27B model, but the cards you pasted above, somehow show that for ELO and TAU2 it's the other way around... Very impressed by unsloth's team releasing the GGUF so quickly, if that's lik…

> Very impressed by unsloth's team releasing the GGUF so quickly, if that's like the qwen 3.5, I'll wait a few more days in case they make a major update. Same here. I can't wait until mlx-community releases MLX optimized versions of these models as well, but happily running the GGUFs in the meantime! Edit: And looks like some of them are up!

absolute n00b here is very confused about the many variations; it looks like the Mac optimized MX versions aren’t available in Ollama yet (I mostly use claude code with this)

Re: Google releases Gemma 4 open models

#380

Post by Wowfunhappy » Fri, Apr 03, 2026, 3:48 AM UTC

car wrote: post #368

Wowfunhappy wrote: post #262

Earlier quoted context omitted.

Hi! Do you ever make quants of the base models? I'm interested in experimenting with them in non-chat contexts.

Yes, they are listed on huggingface. The instruction trained models have an 'it' in their name. https://huggingface.co/collections/unsloth/gemma-4 Edit: Sorry, I'm not sure if this is a quant, but it says 'finetuned' from the Google Gemma 4 parent snapshot. It's the same size as the UD 8-bit quant though.

Only the 'it' models seem to have quants. I was really hoping to try a base model.