More Gemma4 fixes in the past 24 hours

**Reasoning budget fix** (merged): [https://github.com/ggml-org/llama.cpp/pull/21697](https://github.com/ggml-org/llama.cpp/pull/21697)

**New chat templates from Google to fix tool calling:**

31B: [https://huggingface.co/google/gemma-4-31B-it/blob/main/chat\_template.jinja](https://huggingface.co/google/gemma-4-31B-it/blob/main/chat_template.jinja)

27B: [https://huggingface.co/google/gemma-4-26B-A4B-it/blob/main/chat\_template.jinja](https://huggingface.co/google/gemma-4-26B-A4B-it/blob/main/chat_template.jinja)

E4B: [https://huggingface.co/google/gemma-4-E4B-it/blob/main/chat\_template.jinja](https://huggingface.co/google/gemma-4-E4B-it/blob/main/chat_template.jinja)

E2B: [https://huggingface.co/google/gemma-4-E2B-it/blob/main/chat\_template.jinja](https://huggingface.co/google/gemma-4-E2B-it/blob/main/chat_template.jinja)

Please correct me if Im wrong, but you should use these new templates unless you redownload a new GGUF, that has been updated in the past 24 hours with the new template.

You can use specific templates in llama.cpp by the command argument:

--chat-template-file /models/gemma4/gemma4_chat_template_26B.jinja

My current llama-swap/llama.cpp config 26B example (testing on 16GB VRAM , so context window is limited):

"Gemma4-26B-IQ4_XS":

    ttl: 300  # Automatically unloads after 5 mins of inactivity

    cmd: >

      /usr/local/bin/llama-server

       --port ${PORT}

      --host 127.0.0.1

      --model /models/gemma4/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf

      --mmproj /models/gemma4/gemma-4-26B-A4B-it.mmproj-q8_0.gguf

      --chat-template-file /models/gemma4/gemma4_chat_template_26B_09APR2026.jinja

      --cache-type-k q8_0

      --cache-type-v q8_0      

      --n-gpu-layers 99

      --parallel 1

      --batch-size 2048

      --ubatch-size 512

      --ctx-size 16384

      --image-min-tokens 300

      --image-max-tokens 512

      --flash-attn on

      --jinja

      --cache-ram 2048

      -ctxcp 2

    filters:

      stripParams: "temperature, top_p, top_k, min_p, presence_penalty, repeat_penalty"

      setParamsByID:

        "${MODEL_ID}:thinking":

          chat_template_kwargs:

            enable_thinking: true

          reasoning_budget: 4096

          temperature: 1.0

          top_p: 0.95

          top_k: 64

          min_p: 0.0

          presence_penalty: 0.0

          repeat_penalty: 1.0

        "${MODEL_ID}:thinking-coding":

          chat_template_kwargs:

            enable_thinking: true

          reasoning_budget: 4096

          temperature: 1.5

          top_p: 0.95

          top_k: 65

          min_p: 0.0

          presence_penalty: 0.0

          repeat_penalty: 1.0

        "${MODEL_ID}:instruct":

          chat_template_kwargs:

            enable_thinking: false

          temperature: 1.0

          top_p: 0.95

          top_k: 64

          min_p: 0.0

          presence_penalty: 0.0

          repeat_penalty: 1.0"