GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
LessThanThreeAI
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
LessThanThreeAI
Model Tree
LessThanThreeAI/Gemma-4-31B-Humanlike-Chat
Input Modalities
Output Modalities
Supported Functionality
Tested with vLLM 0.31.0 on an RTX PRO 6000 Blackwell (96 GB):
bash
vllm serve LessThanThreeAI/Gemma-4-31B-Humanlike-Chat-GPTQ-Int4 --served-model-name humanlike-gemma \ --max-model-len 16384 --language-model-only \ --enable-auto-tool-choice --tool-call-parser gemma4 --reasoning-parser gemma4Thinking is off by default, as in Google's Gemma 4 template. Send chat_template_kwargs: {"enable_thinking": true} per request to turn it on; the thinking comes back in its own field. It was trained with effort levels: add the line Reasoning effort: low or Reasoning effort: high to your system prompt, or no line for medium. Sampling: temperature 1.0, top_p 0.95, top_k 64 (Google's defaults in generation_config.json). For SGLang, point --model-path at this repo.
GPTQ W4A16: 4-bit symmetric int weights in groups of 128 (static act order) with 16-bit activations, made with llm-compressor 0.14.0.1 in compressed-tensors format. All 410 language linear layers are 4-bit; the vision tower, embeddings and lm_head stay BF16. Calibration: 210 samples, 107,206 tokens of WikiText-2 plus a chat and tool-call mix. Made from the BF16 weights.
The weights take 17.1 GiB in vLLM. On the 96 GB test card at --gpu-memory-utilization 0.88, that left 62.8 GiB for the cache, 74,781 tokens. For a 24 GB card, the IQ4_XS GGUF on the main card runs 32k context in 20.1 GiB.
Same check as every file on the main card: 5 fresh test chats against the reference (the merged BF16 weights in vLLM, which match Google's base plus the Humanlike adapter at runtime: KL 0.0005), served on vLLM 0.31.0, one NVIDIA RTX PRO 6000 Blackwell Server Edition (96 GB).
| File | KL vs reference | Top-1 agreement | Greedy turns identical | Check fails (greedy / sampled) | Decode tok/s |
|---|---|---|---|---|---|
| BF16 | reference | reference | reference | 0 / 0 | 23 |
| FP8 | 0.0034 | 100% | 7 of 15 | 0 / 0 | 40 |
| GPTQ Int4 | 0.0107 | 87.5% | 4 of 15 | 0 / 0 | 67 |
Commissions open, DM codebottle on Discord (https://discord.com/users/320486798859960322). I build custom finetunes like this one: characters, product voices, distillation into smaller models, and domain or use-case specific models.
Apache-2.0, inherited from Google's Gemma 4 release.