Up to date
google Gemma Terms of Use out 13 months ago
268M parameters, text in, text out, 32K tokens of context.
Verified 2026-10-25 Rank #596 in LLMs · score 74.2 method v1 Sources 3 Downloads 30d 16.1k · #472 of 619 tracked
Changed this week: 5 new quantized builds on the Hub (68 in all).2026-10-24 · 4 new quantized builds on the Hub (67 in all).2026-10-24
Picked by rule from 1 tracked piece: the lab's own post, the most shared, a field report, the latest.
The model runs 100% locally in a browser for physics simulation puzzles powered by FunctionGemma and Transformers.js.
Google releases FunctionGemma, a specialized version of the Gemma 3 270M model tuned for function calling to run locally on edge devices.
Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.
No API provider serves it, so its downloads are people running it themselves: 68 quantized builds and 31 Spaces.
Top Spaces running it: webml-community/FunctionGemma-Physics-Playground ♥ 107 · google/functiongemma-tuning-lab ♥ 86 · triflix/brainfuncall ♥ 1 · ovinduG/gemmafunctionC ♥ 1
Weights only, an estimate: leave room for the KV cache. Method
About 0.2 GB at INT4, 0.3 GB at FP8, 0.6 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
Weights, every dated article, and repo commits on one rail.
Teal is the weights (2025-10-08), amber triangles are 1 dated article (each links to it), grey ticks are repo commits.
572 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.
Everything else, folded: the full tables, every source.
| Benchmark | Value | Setting / notes | Reported by | Date |
|---|---|---|---|---|
| author-reported BFCL · Call tools and APIs | 61.6 | 0-shot | model author | 2026-10-07 |
| author-reported BFCL · Call tools and APIs | 39 | 0-shot | model author | 2026-10-07 |
| author-reported BFCL · Call tools and APIs | 36.2 | 0-shot | model author | 2026-10-07 |
| author-reported BFCL · Call tools and APIs | 22.9 | 0-shot | model author | 2026-10-07 |
| author-reported BFCL · Call tools and APIs | 61.1 | 0-shot | model author | 2026-10-07 |
| author-reported BFCL · Call tools and APIs | 73.7 | 0-shot | model author | 2026-10-07 |
| author-reported MultiPL-E · Write code with agents | 63.5 | 0-shot | model author | 2026-10-07 |
| author-reported MultiPL-E · Write code with agents | 29.5 | 0-shot | model author | 2026-10-07 |
| author-reported MultiPL-E · Write code with agents | 25.7 | 0-shot | model author | 2026-10-07 |
| author-reported MultiPL-E · Write code with agents | 20.8 | 0-shot | model author | 2026-10-07 |
FunctionGemma is a 268M parameter text-generation model built on the Gemma 3 architecture.[1][2] It is designed to be fine-tuned for specific function-calling tasks and supports a total input context of 32K tokens.[3][4]
Generated 2026-10-07 15:10 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.
Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.
Listed in LLMs
| Pipeline tag | text-generation |
|---|---|
| Library | transformers |
| Architectures | Gemma3ForCausalLM |
| Model type | gemma3_text |
| Dtype | bfloat16 |
| Params by dtype | BF16: 268,098,176 |
| Size band | S |
| MoE | no |
| Context length | 32,768 (config.json max_position_embeddings) |
| Gated | manual |
| License | Gemma Terms of Use (gemma) |
| Hub snapshot | 2026-10-25 |
vLLM: supported · Gemma3ForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Gemma3ForCausalLM · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed
vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.
Weights-only memory for functiongemma-270m-it: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 0.6 GB · Edge (<4 GB) |
|---|---|
| FP8 (1 byte/param) | 0.3 GB · Edge (<4 GB) |
| INT4 (0.5 bytes/param) | 0.2 GB · Edge (<4 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
finetune: 471quantized: 68adapter: 33
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| unsloth/functiongemma-270m-it | finetune | — | 268M | S | gemma | unsloth | 14 | 104.9k |
| unsloth/functiongemma-270m-it-GGUF | quantized | GGUF | — | — | gemma | unsloth | 98 | 94.4k |
| unsloth/functiongemma-270m-it-unsloth-bnb-4bit | quantized | -bnb | 271M | S | gemma | unsloth | 2 | 37.9k |
| litert-community/functiongemma-270m-ft-mobile-actions | finetune | — | — | — | gemma | litert-community | 299 | 5.7k |
| bartowski/google_functiongemma-270m-it-GGUF | quantized | GGUF | — | — | gemma | bartowski | 3 | 14.7k |
| ggml-org/functiongemma-270m-it-GGUF | quantized | GGUF | — | — | gemma | ggml-org | 6 | 11.8k |
| acon96/Home-FunctionGemma-270m | finetune | GGUF | 268M | S | gemma | acon96 | 12 | 11.3k |
| distil-labs/distil-home-assistant-functiongemma | quantized | GGUF | 268M | S | apache-2.0 | distil-labs | 25 | 8k |
| lmstudio-community/functiongemma-270m-it-MLX-8bit | quantized | MLX | 268M | S | gemma | lmstudio-community | 5 | 8k |
| lmstudio-community/functiongemma-270m-it-MLX-bf16 | finetune | MLX | 268M | S | gemma | lmstudio-community | 0 | 6.3k |
| SkGufranAhmed/functiongemma-finetuned | quantized | GGUF | 268M | S | gemma | SkGufranAhmed | 2 | 4.3k |
| OpenmindAGI/functiongemma-finetuned-g1-multilingual | adapter | — | 268M | S | apache-2.0 | OpenmindAGI | 0 | 4k |
| mlx-community/functiongemma-270m-it-4bit | quantized | -4bit | 268M | S | gemma | mlx-community | 3 | 3.8k |
| mlx-community/functiongemma-270m-it-bf16 | finetune | -bf16 | 268M | S | gemma | mlx-community | 7 | 3.2k |
| onnx-community/functiongemma-270m-it-ONNX | quantized | — | — | — | gemma | onnx-community | 7 | 2.7k |
| DevQuasar/google.functiongemma-270m-it-GGUF | quantized | GGUF | — | — | — | DevQuasar | 0 | 3k |
| DMindAI/DMind-3-nano | finetune | — | — | — | apache-2.0 | DMindAI | 57 | 8 |
| litert-community/FunctionGemma_270M_Mobile_Actions | finetune | — | 268M | S | gemma | litert-community | 12 | 1.2k |
| seyhankokcu/functiongemma-270m-it-simple-tool-calling | finetune | — | 268M | S | — | seyhankokcu | 0 | 1.8k |
| litert-community/functiongemma-270m-ft-tiny-garden | finetune | — | — | — | gemma | litert-community | 24 | 590 |
| BrinqAI/functiongemma-270m-physical-ai | quantized | GGUF | — | — | gemma | BrinqAI | 0 | 1.3k |
| JoaoPito/functiongemma-270m-it-simple-tool-calling | finetune | — | 268M | S | — | JoaoPito | 0 | 1.3k |
| seyhankokcu/functiongemma-270m-it-auraskin-tool-calling | finetune | — | 268M | S | — | seyhankokcu | 0 | 1.3k |
| renhehuang/functiongemma-270m-it-coffee-robot-mcp | quantized | GGUF | 268M | S | gemma | renhehuang | 1 | 1.2k |
| nlouis/ada_model | adapter | — | — | — | — | nlouis | 15 | 388 |
| naver-ellm/functiongemma-270m-it-GGUF | quantized | GGUF | — | — | gemma | naver-ellm | 0 | 1.1k |
| naver-ellm/functiongemma-270m-it-mobile-actions | finetune | — | 268M | S | — | naver-ellm | 0 | 1.1k |
| bhaiyasingh45/functiongemma-function-calling-router_youtube_demo | finetune | — | 268M | S | — | bhaiyasingh45 | 0 | 1k |
| linsong8208/trainer_output | finetune | — | 268M | S | — | linsong8208 | 0 | 1k |
| soniqo/FunctionGemma-270M-LiteRT-LM | adapter | — | — | — | gemma | soniqo | 0 | 991 |
| lmstudio-community/functiongemma-270m-it-MLX-6bit | quantized | MLX | 268M | S | gemma | lmstudio-community | 0 | 904 |
| kartikaybagla/functiongemma-bank-sms-parser | quantized | GGUF | — | — | gemma | kartikaybagla | 0 | 899 |
| Qrzysztof/functiongemma-270m-it-prepaid-cards | finetune | — | 268M | S | — | Qrzysztof | 0 | 888 |
| madhuHuggingface/functiongemma-vpc-gguf | quantized | gguf | — | — | apache-2.0 | madhuHuggingface | 2 | 780 |
| mlx-community/functiongemma-270m-it-8bit | quantized | -8bit | 268M | S | gemma | mlx-community | 1 | 824 |
| zeriworkspace/transfer-function-gemma | quantized | — | — | — | mit | zeriworkspace | 0 | 859 |
| lxyuan/FunctionGemma-270M-banking77-router | finetune | — | 268M | S | gemma | lxyuan | 0 | 824 |
| JackJ1/functiongemma-270m-it-mobile-actions-litertlm | finetune | — | 268M | S | gemma | JackJ1 | 10 | 318 |
| Qrzysztof/functiongemma-270m-it-prepaid-cards-v2 | finetune | — | 268M | S | gemma | Qrzysztof | 0 | 813 |
| ps1x/ha-russian-function-gemma | finetune | — | 268M | S | gemma | ps1x | 5 | 561 |
Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.
Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:39 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:46 UTC · Summary: 2026-10-07 15:10 UTC · Last verified 2026-10-25
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.
